You are on page 1of 37

Hybrid cloud

The New Normal


CEDEXIS EBOOK
Hybrid cloud
The New Normal I Introduction

oday's infrastructure doesn't look like yesterday's. In this Ebook, we look at some of the considerations
Where once Operations teams built extensive, Ops - and especially the newly-minted DevOps - teams
high-availability data centers, today they extend their must keep in mind as they integrate public cloud ser-
capabilities with the elastic storage, compute, and deli- vices into their hybrid cloud environments.
very capacity of public cloud services. From the serious - (identifying the hidden risks lurking in
hybrid infrastructures), to the quirky (a lighthearted look
But public cloud infrastructure comes at a cost. at how to make the next emergency more entertaining),
Not only must Ops teams manage the chaos of depart- this collection of perspectives provides a window into
ments throughout the organization grabbing new cloud the challenging world of the hybrid cloud operation.
capacity at will, they must somehow ensure that they
balance delivering excellent experience to each and Hybrid cloud is the new normal - but that doesn't mean
every end user with cost structures that can make or it's all worked out just yet.
break the profit/loss equation of the whole business.

CEDEXIS EBOOK - OCTOBER 2017 02


Hybrid cloud
The New Normal I Table of contents

The problem lurking Self-sourced data won’t get you


inside your hybrid IT strategy 05 where you want to go 22

SaaS Metrics that beat


comes at you fast 09 Murphy’s law 26

10 ways to make Cloud-first + DevOps-first


outage mitigation fun 14 = 81% competitive advantage 30

Cloud capabilities turn developers Optimizing for resources


into operational experts 18 and consumers alike 33

CEDEXIS EBOOK - OCTOBER 2017 03


Hybrid cloud
The New Normal I The authors

Charles Araujo is the accidental Andrew Marshall has driven pro- Simon Jones is a Silicon Valley
author of the best selling book, duct marketing, launch, and GTM veteran, who has spent the last two
The Quantum Age of IT: Why strategy for companies like decades building, evangelizing, and
Everything You Know About IT is Amazon Web Services, Microsoft, implementing technology solutions
About to Change. He is a regular and New Relic. His passion for new that are often sufficiently advanced
contributor to both CIO Insight and disruptive SaaS-based tools led that they are virtually indistingui-
Magazine and InformationWeek him to Cedexis, where he is the shable from magic.
and has been quoted in or publi- Director of Product Marketing. Currently Evangelist and head of
shed in magazines and websites He is a contributor to APM Digest, marketing for Cedexis, he resides at
including Time, CIO, CIO & Leader, DevOps Digest, DevOps.com, and the heart of the San Francisco Bay
IT Business Edge, TechRepublic, other technology publications. Area, and spend every day
Computerworld, USA Today and Andrew lives, works, and plays in endeavoring to explain compli-
Forbes. He currently also serves as the Silicon Forrest, but he prefers cated things in simple ways.
Principal Analyst for to just call it Portland.
Intellyx.

CEDEXIS EBOOK - OCTOBER 2017 04


It is tempting to dismiss
something that you see as
trendy. While it implies pro-
gressiveness and
forward-thinking and im-
parts a certain cachet on
the trendsetter, observers
often see those things
deemed trendy as mere
fads. Enterprise executives
The problem lurking are just as subject to the
inside your hybrid IT strategy allure of the latest trend —
and today’s object of fasci-
Charles Araujo
BrainBlog, August 30, 2017 nation is Hybrid IT.

HYBRID CLOUD - THE NEW NORMAL 05


Unlike many trends that are, in fact, nothing but fads, Will the real hybrid IT please stand up
Hybrid IT is the real deal. It represents the natural evolution
and rationalization of a number of trends and movements The reason that many enterprise leaders do not yet unders-

that have consumed enterprise IT for the last decade. tand the problems they may soon face in routing traffi-
cacross their Hybrid IT architectures is that many of them

More importantly, it embodies a movement away from the do not yet fully understand the real nature of a Hybrid IT

initial period of exploration and experimentation to a time strategy.

in which organizations embrace emerging technologies at


scale and begin the arduous task of rationalizing and inte- As I wrote recently, we define Hybrid IT as “a management

grating their legacy technology investments with the approach in which organizations create a workload-centric

modern technologies and strategies necessary to thrive in and value-driven integrated technology stack that may

the digital era. include legacy infrastructure, web-scale architectures,


private cloud implementations along with public cloud

As enterprise executives embrace the ideas behind Hybrid platforms ranging from Infrastructure-as-a-Service to Sof-

IT, they must also face the fact that there is a significant tware-as-a-Service.”

problem lurking within their evolving strategies: the inability


to effectively, predictively and cost-optimally distribute The problem is that many enterprise organizations are

traffic across their application and rich media content. using the Hybrid IT trend as a crutch, skipping over the
workload-centric and value-driven parts and taking the

It’s a problem that they must solve to realize their Hybrid IT concept to mean that they can pick-and-choose technolo-

ambitions. gies indiscriminately. With that approach, traffic is less of a


challenge as they have done nothing more than bolt-to-
gether a mishmash of technology pieces and parts.

As organizations embrace the real meaning of a Hybrid IT


strategy, however, and, as a result, reimagine their architec-

THE PROBLEM LURKING INSIDE YOUR HYBRID IT STRATEGY 06


ture from a workload and business-value perspective, they applications and content whenever and wherever cus-
will find that traffic flows change. And as they change, the tomers demand it. As organizations continue to shift away
traffic distribution problem lurking in the shadows will rear from systems-drive architectures and embrace worklo-
its ugly head. ad-centricity, the challenge of managing and allocating
traffic will become increasingly critical to enhancing the
customer experience and protecting competitive value for
Traffic distribution in a hybrid world the organization. Those organizations that can overcome
this challenge will reap the rewards.
In traditional architectures, traffic patterns are relatively
predictable and manageable. In cloud deployments, cloud
providers largely handle the management of traffic.
The Intellyx take
Hybrid IT architectures, with their workload-centricity,
however, introduce a complexity that most enterprise
organizations are not yet prepared to handle. As organiza-
tions dynamically allocate application workloads and sup-
porting content between on-premises infrastructure and
one or more cloud environments, traffic management
becomes nearly impossible using traditional traffic optimi-
zation approaches.

Moreover, rapidly changing customer expectations and


what we at Intellyx call the primacy of the customer, mean
that enterprise organizations are no longer in control of Workload-centricity and the adoption of authentic Hybrid IT
traffic distribution at all — at least not in the traditional strategies, like everything related to digital transformation, will
sense. Instead, organizations must be capable of delivering require fundamental shifts in both culture and technology.

THE PROBLEM LURKING INSIDE YOUR HYBRID IT STRATEGY 07


Organizations must first change the way they think about
how they deploy applications and content across their
infrastructure and the implications of traffic patterns on the
customer experience. As they do so, they must also re-eva-
luate the technologies that they rely on to manage the
delivery of those applications and content.

This re-evaluation will lead enterprise leaders to explore


emerging technology companies, such as Cedexis, that
deliver both visibility and software-defined application
delivery capabilities. The ability of such tools to dynamically
and automatically route traffic based on business logic and
optimization rules will be critical to organizations seeking to
adopt Hybrid IT strategies.

Adopting a Hybrid IT strategy, however, is not an end unto


itself. As we examine in the webinar, This is How to Make
Hybrid IT Web-Scale, the ultimate goal of such a strategy is
to enable organizations to embrace the web-scale ethos
needed to compete in the future — including the dynamic
management of traffic — while grappling with the architec-
tural reality of the modern enterprise.

Copyright © Intellyx LLC. Cedexis is an Intellyx client. Intellyx retains full editorial
control over the content of this paper.

THE PROBLEM LURKING INSIDE YOUR HYBRID IT STRATEGY 08


SaaS comes at you fast
These days, it seems only a software-defined
application delivery platform can keep up with
the needs of multi-cloud and dynamic
infrastructure.

Andrew Marshall
DZone / Cloud Zone, August 31, 2017

HYBRID CLOUD - THE NEW NORMAL 09


It’s an exciting time to be software! Chances are you were by the end user, and easily write contingency-based code
brought into the world through continuous delivery, to implement those rules.
making a loving team of DevOps parents very proud… and Changes to your application architecture aren’t a major
very, very happy. concern since your rules can accommodate both planned
and unexpected changes.
Your potential is unlimited! The only thing that could pos-
sibly prevent you from growing up to fulfill your SaaS desti-
ny is a poor application delivery strategy.
Your generation is right at home on
heterogeneous infrastructure
Like applications, modern infrastructure has also changed
Your grandfather was a monolithic
significantly in the last several years. No longer are you
app using a single data center, or perhaps a data center and
In a simpler time, maybe five or six years ago, applications one CDN to make sure your app gets to your customers.
were monolithic in nature, making delivery a straight Multi-CDN infrastructures allow you to re-route app traffic
forward process. That’s no longer the case, as today’s when performance is an issue.
modern apps are distributed and often powered by micro-
services like containers or cloud services. Multi-cloud environments are becoming more and more
prevalent as enterprises want to avoid public cloud vendor
DevOps teams these days need to make sure every com- lock-in and gain more control over their critical shared
ponent of the application is performing as expected. infrastructure.
But that’s just the start, as they also must make sure the
connective tissue of these distributed apps isn't underper- And of course, the data center you purchased a few years
forming, which could create a terrible end-user experience. ago isn’t just gathering dust…you paid for it, so you’re using
Software-defined application delivery allows you to decide it. The net result is a heterogeneous set of infrastructures
exactly how you want your application to be experienced being used to deliver you to customers.

SAAS COMES AT YOU FAST 10


Kids today with their dynamic in- Do you really want to do that manually, while in the middle
of a migration? Software-defined application delivery
frastructure systems make sure your app reaches customers at all
Even if your infrastructure is fully built out and provisioned, times.
it is, by the nature of the cloud, constantly changing as
instances are both spun-up and de-provisioned.
Often, and for many reasons, this happens without you
Learn to speak a multi-cloud foreign
explicitly knowing about it. Software-defined application
delivery helps you optimize your infrastructure in its current
language
state, again using the business rules you decide on. There are plenty of reasons to shift to a multi-cloud
This has the additional benefit of giving you a better idea of infrastructure plan: avoiding cloud lock-in, protection
how to plan for cloud capacity and avoid overpaying since against massive cloud outages, leveraging certain ven-
you can automate delivery based on your reserved dor-specific services or offerings, and good old fashion
instances and fully utilize the capacity you paid for. Cloud SLA negotiation leverage.
Your Software-defined application delivery also accounts But once you’ve made the decision to add a second cloud
for changes in cloud provider pricing and specific geogra- vendor to your infrastructure, it’s time to start thinking
phic needs, keeping the end of month bill surprises to a about how to ensure your app delivery is optimized for
minimum. multi-cloud environments. Having your Ops team do so
manually with each vendor isn’t likely to be a popular strate-
gy. The only way to manage delivery is with a software-de-
fined abstraction layer on top of both clouds.
Moving sucks Your Software-defined application delivery speaks both
There aren’t any shortcuts when it comes to migrating appli- cloud “languages”, and does all the translating for you,
cations to the cloud. The first step in cloud migration needs automatically.
to be making sure your application can be delivered to your
worldwide customers before, during, and after the migration.

SAAS COMES AT YOU FAST 11


Your grandfather’s application delive- Software-defined application delivery automation has the
very practical effect of removing many common delivery
ry controllers can’t keep up issues, such as micro-outages, service degradations, and
Remember the monolithic app? Its best friend was the the occasional (and highly newsworthy) public cloud
Application Delivery Controller (ADC), also built for a diffe- outage from the “worry” list. Your uptime and customer
rent age. Often built into a blinking hardware box in a data experience should all be ensured by a set of routing rules
center, old guard ADCs had several drawbacks. you determine ahead of time. You can’t put a price on
They required a significant upfront expense, as opposed peace of mind, or not getting a million “OUR APP IS
today’s cloud pay as you go model. Monitoring was done DOWN!!!!!” emails.
with synthetic testing only, not the combination of real-user
and synthetic monitoring many software-defined applica-
tion delivery solutions provide today. SaaS comes at you fast
Most importantly, ADC solutions were not as easily confi- Outages and cloud service degradation happens, so you
gurable and customizable as the robust software-defined need a plan in place for when they do. Many companies
application delivery, making them unusable with today’s choose to utilize multiple cloud providers to ensure their
modern application architecture. application is always available to worldwide customers,
even during an outage.

Taking control over changes in public cloud service offe-


Get in a routine rings, pricing models, and SLAs is another powerful motiva-
As more and more concerns (security, DevOps, new tech- tor for Ops teams to move to a multi-cloud architecture.
nologies, etc.) are added to the Ops team’s list of things to Software-defined application delivery allows you to
worry about, the need for process automation becomes seamlessly and automatically deliver applications over
paramount. multiple clouds and/or CDNs, factoring in real-time availa-
bility, costs structures, and application performance.

SAAS COMES AT YOU FAST 12


There's no fate but what we make
for ourselves
There’s no way you can keep up with constantly changing
applications, the infrastructure they’re built on, and inevi-
table network issues with manual processes.
Software-defined application delivery allows you to imple-
ment the business rules that you know are important to the
app delivery process, and do so instantly and automatically
via customizable code.

End user experience, network latency, unplanned cloud


provisioning or CDN bursting, and other considerations can
all be factored in as your needs change.
The multiple application delivery fail-safe layers will keep
you insulated from the outages that need to be mitigated
during your Game of Thrones viewing.

SAAS COMES AT YOU FAST 13


It’s 3:47 a.m. You and the
rest of the Ops team have
been summoned from
your peaceful slumber to
mitigate an application de-
livery outage.
As you catch up on the
frantic emails, Slack chats
and text messages from
everyone on your interna-
tional sales team (Every.
Single. One.), your mind
races as you switch to pro-
10 ways to make blem-solving mode.
outage mitigation fun It’s time to start thinking
about how to make this
Andrew Marshall
DevOps.com, August 25, 2017 mitigation FUN!

HYBRID CLOUD - THE NEW NORMAL 14


1. vent their creative energy (and frustration) with some fun
customized MS Paint updates to the offending page.
Gently remind everyone that the SaaS-delivered automated
Bonus points for art that reminds everyone of the value of a
application delivery strategy you proposed would have
multi-cloud strategy powered by a programmable applica-
prevented this. No need to rub it in, but … if you had turned
tion delivery platform.
on a software-defined application delivery platform for your
hybrid infrastructure, you’d all be sound asleep right now.
Automated real-time delivery decisions and failover would
be nice, right? Just sayin’. 4.
Use health metrics to cure ills instead of just conducting
another really good post-mortem. Tired of “learning
2. lessons” from these emergency room drills?
Drop a generationally divisive Giphy in the emergency Slack You depend on NGINX for your LLB, but don’t have a way
channel. Your coworkers’ opinions on gaming consoles, to use those LLB health metrics and data to automate
Spiderman movies and music are different than yours! global delivery. Disjointed delivery intelligence means you
The perfect animated GIF will remind them of your erudite don’t know how your apps will land with users.
tastes, while you have their attention. You need an end-user-centric approach to app delivery that
Extra credit if you can work in some low-key shade about automates the best delivery path and ensures failover is in
not listening to your equally sophisticated opinions on place at all times.
optimized outage mitigation.
Micro-outages often fly under your passive monitoring
radar, but that doesn’t mean your users don’t notice them.
3. An active, integrated app delivery approach re-routes auto-
Hold an impromptu cloud health dashboard page customi- matically before you lose business.
zation contest. Oh hey, look at that! Your cloud provider’s Post-mortems are fun … but so is making sure your apps
health dashboard page says everything is fine … because it’s survive the last mile. Arguably more so?
powered by the services that went down. Help your team

10 WAYS TO MAKE OUTAGE MITIGATION FUN 15


5. 7.
Open (and then close) some tickets for the sales team. Leverage real user measurements (RUM) to foresee and fix
“Sales needs to sell more stuff.” You’ll feel better. problems before they happen. Probably best to do this
during normal work hours.
User experience data from around the world can detect
6. degrading, sluggish resources in real time and user-centric
Build a countdown clock for when you can finally replace app delivery logic powered by RUM can make quick rerou-
your old ADC hardware. Sure, your Mode1 ADC hardware ting decisions automatically.
was a sunk cost so you’re stuck with them for a while. No more getting woken up after the application crashes.
But you’re one unnecessary emergency closer to having a While you’re wishing you had RUM on your side, you can
fully software-defined application delivery platform for your look up some fun facts from the countries experiencing
hybrid cloud. And now you’re even closer. And closer … app outages. Did you know Luxembourgish is an official
Tick. Tock. language?

10 WAYS TO MAKE OUTAGE MITIGATION FUN 16


8.
Send an email. You, too, can be the weird colleague who
sends emails with crazy, middle-of-the-night time stamps.

9.
Do some virtual window shopping. Browse around to see
what you could have purchased with the money you were
just forced to spend on unplanned cloud instance provisio-
ning to keep your app running. That desktop Nerf missile
launcher (or 700 of them) would have been pretty nice.

10.
Swear off the game of chance. You’ve just proven it’s not
that much fun after all. Don’t just dump everything onto
one cloud and call it done.
Clouds go down for so many reasons. Use an application
delivery platform control layer to build in the capability to
auto-switch to an available

10 WAYS TO MAKE OUTAGE MITIGATION FUN 17


Crowdsourced intelligence
is a powerful concept,
brought to reality and
fueled by the Internet and
mobile devices. Without
the communications
technology and vast but
addressable audience of
smartphone addicts,
Self-sourced data marketing projects

won’t get you where dependent on gathering


data or pooling microtasks
you want to go from an undefined,

Simon Jones
unrelated audience would
Cedexis blog, August 31, 2017 remain mostly theoretical.

HYBRID CLOUD - THE NEW NORMAL 22


Of course, companies outsourced tasks and polled cus- never driven before. Do you want to rely on your individual
tomers for data before crowdsourcing came along, but experience, or do you want to consult the Waze app and
soliciting a thousand survey responses takes a surprising figure out how to avoid gridlock? (Bonus points for having
amount of work – and, importantly, time – without the enough time for a side trip to Starbucks.)
mobile Internet. Massive data projects like the Census or
NIH-funded longitudinal studies were carried out long Amazingly, we’re now able to apply the power of crowd-
before the Internet, but required complex coordination, sourcing to networks, endpoints, and bytes just as we to
significant resources, and a long timeline. people and cars.

But ever since the term “crowdsource” was coined more In the Zettabyte era, community-powered intelligence is a
than a decade ago, the rapid, large-scale assembling of fundamental need if we’re really going to move 20,000
ideas, opinions, funds, collaborators, computing power and Gbps of global Internet traffic without constant outages,
labor online has become commonplace. In its myriad, still slowdowns, and the inefficient yet miraculous worka-
evolving forms, crowdsourcing has proven to be one of the rounds we too often demand from Ops teams — while also
most groundbreaking tools of the mobile era. controlling costs. Unless you are in the very highest eche-
lon of traffic delivery, your service simply won’t create
Crowdsourcing technology has moved well beyond enough data to truly map the Internet. Heck, even if you
market research and fundraising. One of the most popular got 100 million hits a day, the chances are good they’d all
examples is Waze, where driver experience reports, be clustered in a handful of major regions.
construction plans, emergency dispatches, and traffic
metrics are combined with algorithms to produce real-time If you expand into a new market, or your app suddenly
guidance for drivers. You may drive around the Bay Area a becomes popular in Saskatchewan, or for any number of
lot and have a strong sense of traffic patterns based on reasons traffic from a previously quiet region surges
experience. But what happens if there’s an unexpected suddenly (consider the stochastic outcomes of natural
event— let’s say a tornado in Palo Alto—or you have two disasters, widespread cyber attacks, and political and cele-
hours to get to a meeting near Sacramento, where you’ve brity happenings), you won’t have enough visibility or intelli-

SELF-SOURCED DATA WON’T GET YOU WHERE YOU WANT TO GO 23


gence on hand to manage your traffic in that particular – Banding these services together makes it possible to see
suddenly vital! – corner of the Internet. You need to essentially all the networks in the world each day. We’re
crowdsource the data that will give you the comprehensive talking about all the “little Internets ”Radar collects data from
view of the Internet you need to avoid outages, ensure high more than 50,000 of these networks daily. More than 130
quality of experience (QoE) for users, and make efficient major service providers feed metrics into the system each
use of your resources. day. All told, hundreds of millions of clients generate over
14 billion RUM data points every day. That’s quite a crowd –
Let’ use Cedexis’ real user monitoring community Radar to and one that basically no service could pull together all on
illustrate the challenge you’re up against. Our data sets are its own.
based on Real User Monitoring (RUM) of Internet perfor-
mance at every step between the client and the clouds, Community doesn’t just give you breadth of course, but
data centers, or CDNs hosting your application. To source also depth: we get at least 10,000 individual data points
this data, we have created the world’s largest user expe- each day from over 13,000 of those ASNs. You simply can’t
rience community, in which publishers large and small glean that kind of traffic intelligence from your own system.
deposit network observations from their own service, then Do the math: you’ll see that even if you have 100 million
share the aggregate total, so they can benefit from one visitors each day, likely less than half are coming from
another’s presence. outside the major pockets (concentrated ASNs), leaving
you very few data points spread across the thousands of
remaining ASNs. So when the first visitor of the day turns up
from Des Moines, Iowa, or Ulan Bator, or Paris, France,
you’d have no data handy to make intelligent traffic deci-
50,000 130 14 billion sions.
networks service providers RUM data points
Radar collects data from across Over 130 service providers Hundreds of millions of clients Everyone needs community intelligence. Not just for the
more than 50,000 networks feed metrics into Radar generate over 14 billion RUM
daily daily data points every day newest, least understood pockets of users on the edges of
your network. Crowdsourced community intelligence from

SELF-SOURCED DATA WON’T GET YOU WHERE YOU WANT TO GO 24


the big, messy pools of so provides the early warning Instead, rely on community-based telemetry to produce
system every Operations team needs to keep the wheels the actionable intelligence and predictive power you need
turning, and the user using. to serve your end-users, no matter where they pop up.
Many countries have thousands of ISPs. In Brazil, the Radar
community supplies 10,000 daily data points from 1,595 With crowdsourced data sets from the global Internet
different ASNs. Russia, Canada, Australia, and Argentina also community, you already have instant access to intelligence
have enormous diversity of ISPs, especially in relation to about QoE, bottlenecks, and slowdowns — before you
their populations. These locations are likely to be central to even know you need it.
your business success and global content delivery needs.
Having user experience data of this breadth and depth is
particularly important where there are so many individual
peering agreements and technical relationships, represen-
ting countless causes for variable performance.
When there are countless ways for application delivery to
go wrong, you need granular data to feed into intelligent
delivery algorithms to ensure that, in the end, everything
goes right.

When you’re trying to manage application and content


delivery globally, you need visibility into thousands of ASNs
on a continuous basis. Unless you’re Google, you’re not
going to touch most of these networks at the depth you
need. So unless you have a really cool crystal ball, you have
no idea how they are performing. You’d have to conduct
an enormous amount of traffic to gain a comprehensive,
real-time view on your own.

SELF-SOURCED DATA WON’T GET YOU WHERE YOU WANT TO GO 25


The upside-down world of
digital transformation can
leave you feeling like you’ve
fallen down an internet
rabbit hole.
If you’re a DevOps denizen,
the swirling pace and scale
of virtualization might give
you that “lost in the Techni-
color forest” feeling when
it’s time to release an app
Cloud capabilities turn into the wild. “Curiouser

developers into operational and curiouser!” you say as


you try to sort through
experts hybrid combinations of
private and public clouds,
Andrew Marshall
DevOps.com, August 7, 2017 CDNs, and data centers.

HYBRID CLOUD - THE NEW NORMAL 18


As a developer, you worked hard on the design, build and As DevOps teams focus more on microservices and contai-
test—now you want to make sure it takes the best, fastest ners and less on “stacking and racking” servers, perfo-
path to microservices and users. It’s folly to trust the answer mance, control and delivery become the focus areas that
from any one source—synthetic measurements may purr matter. As the division between Dev and Ops experts conti-
like a Cheshire Cat over hardware health, meanwhile the nues to dissolve, each becomes increasingly aware of the
White Rabbit of RUM (real user measurements) is spazzing other’s roles, responsibilities and skill sets.
out over latency and traffic jams. The goals of agile development, non-stop testing and
The lesson Alice finally learns in Lewis’ timeless tale? optimized delivery become the shared destination; paths
She must learn to control the journey herself. merge until everyone is creating momentum around the
prized feedback loop. As infrastructure and services
become more abstracted, the focus is less on writing origi-
nal code and building stacks and more on selecting
Developing for the complex cloud
resources.
Fortunately, there’s a way for developers to mix that “just
right” solution that will reveal the right path through the The selection and use of these resources has to be strategi-
internet forest. This used to be the purview of opera- cally guided by real-time intelligence to meet multiple
tions—setting up servers and delivery conduits. performance and business objectives.
In the pre-cloud era, developers just wrote code. Somebo- The more abstract and hybridized the infrastructure beco-
dy in Ops pushed it live, and that part of the process was mes—and it is moving rapidly in that direction for every
opaque to the developer. With the proliferation and matura- business—the harder it is to navigate.
tion of the cloud, the Ops team is increasingly focused on
dynamic infrastructure and services. Developers routinely If you can’t keep all your resources straight, you can’t opti-
spin up their own servers. mize them. Centralizing and automating tracking and traffic
management is mission-critical.

CLOUD CAPABILITIES TURN DEVELOPERS INTO OPERATIONAL EXPERTS 19


The delivery is in the code Developers are now actively part of the process that brings
together global and local traffic management data (RUM
and infrastructure health), application performance mana-
gement and cost considerations to ensure the cheapest,
fastest path to that microservice. The balance and mix of
these factors can be customized and fine-tuned by the
developer (a.k.a., the newly minted operations expert).

Which brings us back to the developer’s mandate to


The bleeding edge becomes the be-
control and optimize delivery from the application. drock
Cloud-based delivery tools combine real-time user mea- You know the cloud is no fairy tale when 74 percent of tech
surement with scriptable application traffic shaping and CFOs claim that this year the most measurable impact on
built-in cost controls. Gone are the days of blaming poor their business will come directly from cloud computing.
performance on the network. End users expect seamless, Spending on cloud computing is set to grow at 19 percent
instantaneous experiences, no matter what. This is a tall CAGR through 2020, six times the rate of overall IT spen-
order for modern apps that combine video, microservices ding. Developers who can harness the magic of hybrid
and dynamic content. To ensure high quality of experience cloud/CDN for video, web content and application delivery
(QoE), you have to automate and optimize with will be rock stars. Picture it: fully automated delivery based
software-defined delivery. Developers can now contribute on business rules, actionable data from real user and syn-
to optimization by coding in delivery intelligence. thetic testing, and self-healing network optimizations.
Developers can push their capabilities and control right to
This isn’t merely another task for the developer; it’s an the bleeding edge, making them an even more agile contri-
opportunity. Developers are now empowered to deliver the butor to the continuous improvement cycle.
functionality of their product or service exactly as designed.

CLOUD CAPABILITIES TURN DEVELOPERS INTO OPERATIONAL EXPERTS 20


Soon enough, as technology progresses and infrastructure
is increasingly software-defined, these skills will become a
requirement.

Over the next few years, this data-driven approach to app


and content delivery will be considered a standard capabi-
lity for digital businesses and a core responsibility for deve-
lopers. The forward momentum of organizations that have
committed to both cloud-first and DevOps-first strategies is
powerful and will quickly boost early adopting competitors
out of reach.

The number of developer-operators capable of managing


application delivery in capacity-aware, consumer-res-
ponsive, application-tuned ways will soon build to critical
mass. Don’t get left behind, wandering lost in the woods.

CLOUD CAPABILITIES TURN DEVELOPERS INTO OPERATIONAL EXPERTS 21


Automobile congestion is
a scourge in most cities
around the world.
The negative impacts are
varied: personal time
wasted sitting in traffic,
Metrics that beat dangerous accidents,
delayed arrival of
Murphy’s law emergency vehicles,
Simon Jones environmental damage
Cedexis blog, August 23, 2017 from stop-and-go driving,
inefficient delivery
logistics… the list goes on.

HYBRID CLOUD - THE NEW NORMAL 26


Stop signs and simple on-off timed signals were sufficient Internet infrastructure — increasingly a combination of data
to direct traffic in the early years; in recent decades, camera centers, CDNs, co-los, and regional AWS resources — we
and sensor-triggered traffic lights became necessary to need an innovative smart traffic management system, too.
address mounting congestion and protect pedestrians. Like the street traffic control systems in widespread use
With congestion and pollution problems mounting, Ligh- today, basic Global Load Balancers (GLBs) are an essential
thouse Cities like Cologne, Barcelona and Stockholm are start, but we need something more dynamically
piloting smart traffic management systems. These innova- data-driven, real-time responsive, and configurable to spe-
tive combinations of software controllers and connected cific scenarios. This is especially true for DevOps organiza-
sensors in traffic lights and vehicles will optimize travel tions.
speeds, delivery routes, and public transportation links to
prevent accidents, untangle bottlenecks, and balance the
ratio of car drivers and rail riders. Murphy’s law applies:
All the things that can go wrong…
When LLBs are not intelligent and dynamically controlled,
Building towards a future of sustai-
quality of experience (QoE) degrades, provisioning econo-
nable internet traffic mics are out of whack, and outages occur. Even with failo-
Internet traffic has followed similar patterns on an accele- ver mechanisms in place, if traffic gets sent to an LLB clus-
rated timeline. Local load balancers were sufficient to divvy ter that is down, it often takes an unacceptable amount of
up incoming requests between a handful of servers sitting time to switch over. Sometimes, a location is working fine
in the same place. With the explosion of Internet data use but starting to bump up against resource limits.
and development of cloud and virtual infrastructure, local A shift or failure in another location could cause so much
load balancers (LLBs) have proliferated and need to be traffic to flow to the near-capacity resource that it breaks,
managed like endpoints themselves. Application Delivery resulting in an entirely avoidable outage (and another fun
Controllers address this issue for data centers, but aren’t post-midnight emergency intervention).
upto the task in a hybrid IT world. To optimize modern In multi-cloud and hybrid infrastructure for application and

METRICS THAT BEAT MURPHY’S LAW 27


media delivery, the sought-after advantages of scalability,
agility, and affordability are lost when there is no ove-
rarching control layer. Without intelligent global traffic
Outages are unnecessary evils
control, one location will be overused and in danger of Everybody hates outages – users, developers, sys admins,
failing if something unexpected happens, while another sales teams, and executives. You might be able to keep
location will be underused. Data Center resources are your cloud budget woes to yourself, but outages big and
already paid for under CapEx, as opposed to OpEx cloud small chip away at your brand, reputation, and app or site
spot instances; intelligent, configurable GLBs can maintain popularity.
the difficult balance between low cost and high cost loca-
tions in your specific resource mix. Dynamic GLBs use real-time health checks to detect
potential traffic or resource problems, route around them,
and send an alert before failure occurs so that you can
address the root cause (during normal work hours, imagine
that).

Even with a failover plan, LLBs allowed to run without


real-time intelligence are susceptible to slowdowns,
micro-outages, and cascading failures, especially if hit with
a DDOS attack or unexpected surge.

There are times when it’s necessary to shift your standard


resource model: updates, repairs, natural disasters, and app
or service launches. Without scriptable load balancing, you
have to dedicate significant time to shifting resources
around — and problems mount quickly if someone takes
down a resource but forgets to make the proper notifica-
tions and preparations ahead of time.

METRICS THAT BEAT MURPHY’S LAW 28


designed to work the same way on all platforms, so
Intelligent GLBs are here to save smooth deployment is possible, no matter your resource
mix.
the day
The direct benefits of implementing a scriptable, It may be hard to imagine now, especially if you are stuck in
user-configurable, data-driven global load balancing plat- the transportation grind of a major urban area, but we may
form for hybrid architectures are three-fold: performance, one day live in cities where cars and public transportation
economics, and control (including the ability to account for flow and connect seamlessly, air pollution is minimal, and
region-specific regulatory requirements). road rage is an embarrassing relic of the past.

Feeding high quality, real-time datasets (LLB and cloud When it comes to Internet traffic, we can start living the
monitoring, server availability, real user measurements, and dream much sooner. After all, nobody in 1995 would have
other resource health checks) into traffic decision engines believed predictions about binge watching entire sitcom
automates the entire delivery path, optimizing application seasons on your handheld wireless computer while going
availability and latency for consistently high QoE, swift about your daily routine. With intelligent, dynamic global
resolution of congestion and outages, and fine-tuned load balancing, we’re well on our way to yet another step
control over resource use and cost. change in application and content delivery.

This automated approach to software-defined application


delivery is essential for DevOps innovation; continuous
integration and delivery methods can’t wait for hardware
changes, and the growing use of microservices and contai-
ners requires application-level control that developers can
understand and configure.

Moreover, software-defined global load balancers are

METRICS THAT BEAT MURPHY’S LAW 29


Cloud-First
+ DevOps-First
= 81% Competitive Advantage
To ensure that an app performs as designed, and end users
have a high quality experience, agile teams need to
automate and optimize with software-defined delivery.

Simon Jones
Cedexis blog, August 23, 2017

HYBRID CLOUD - THE NEW NORMAL 30


We recently ran across a fascinating article by Jason category than the DevOps-only companies.
Bloomberg, a recognized expert on agile digital transforma-
tion, that examines the interplay between Cloud-First and These findings, among others, got us to thinking about the
DevOps-First models. That article led us, in turn, to an potential benefits and advantages such Delivery Disruptors
infographic centered on some remarkable findings from a can gain from adding Cedexis solutions into their powerful
CA Technologies survey of 900-plus IT pros from around mix. Say, for example, you have agile dev teams working on
the world. The survey set out to explore the synergies new products and apps and you want to shorten the exe-
between Cloud and DevOps, specifically in regards to cution time for new cloud projects.
software delivery. You can probably guess why we snapped
to attention. To let your developers focus on writing code, you need an
app delivery control layer that supports multiple teams and
The study found that 20 percent of the organizations repre- architectures.
sented identified themselves as being strongly committed
to both Cloud and DevOps, and their software delivery
outperformed other categories (Cloud only, DevOps only,
and slow adopters) by 81 percent.

This group earned the label “Delivery Disruptors” for their


outlying success at maximizing agility and velocity on
software projects. On factors of predictability, quality, user
experience, and cost control, the Disruptor organizations
soared above those employing traditional methods, as well
as Cloud-only and DevOps-only methods, by large percen-
tages. With the Cedexis application delivery platform, you can
For example, Delivery Disruptors were 117 percent better at support agile processes, deliver frequent releases, control
cost control than Slow Movers, and 75 percent better in this cloud and CDN costs, guarantee uptime and performance,

CLOUD-FIRST+ DEVOPS-FIRST = 81% COMPETITIVE ADVANTAGE 00


31
and optimize hybrid infrastructure. Your teams get to work user and synthetic testing, and self-healing network optimi-
their way, in their specific environment, without worrying zations. Incorporating these capabilities with a maturing
about delivery issues looming around every corner. Cloud-first and DevOps-first approach will likely put the top
performers so far ahead of the rest of the pack, they’ll
Application development is constantly changing thanks to barely be on the same racetrack.
advances like containerization and microservice architec-
ture — not to mention escalating consumer demand for
seamless functionality and instant rich media content. And
in a hybrid, multi-cloud era, infrastructure is so complex
and abstracted, delivery intelligence has to be embedded in
the application (you can read more about what our Archi-
tect, Josh Gray, has to say about delivery-as-code here).

To ensure that an app performs as designed, and end users


have a high quality experience, agile teams need to auto-
mate and optimize with software-defined delivery. Agile
teams can achieve new levels of delivery disruption by
bringing together global and local traffic management data
(for instance, RUM, synthetic monitoring results, and local
load balancer health), application performance manage-
ment, and cost considerations to ensure the optimal path
through datacenters, clouds, and CDNs.

Imagine the agility and speed a team can contribute to


digital transformation initiatives with fully automated app
delivery based on business rules, actionable data from real

CLOUD-FIRST+ DEVOPS-FIRST = 81% COMPETITIVE ADVANTAGE 32


One of the genuinely diffi-
cult decisions being made
by DevOps, Ops, and even
straight-up developers
today is how to ensure
outstanding quality of ex-
perience (QoE) for their
users.
Do you balance the hard-
ware (physical, virtual, or
otherwise) for optimal
Optimizing load? Or track quality of
for resources service (QoS) metrics – like
throughput, latency, video
and consumers alike start time, and so forth –
and use those as the pri-
Simon Jones
Cedexis blog, June 20, 2017 mary guide?

HYBRID CLOUD - THE NEW NORMAL 33


It’s really not a great choice, which is why we’re happy to Think of it as a marriage of local traffic management (LTM)
say, the right answer to the question of whether to use and global traffic management (GTM). LTM takes care of
local or global traffic management is: both. routing traffic that arrives at a (virtual or physical) location
between individual resources efficiently, ensuring that
It hasn’t been a great choice in the past because, while resources don’t get overloaded (and of course, spins up
synthetic and real user measurements (RUM) overlap pretty new instances when ready); GTM takes care of working out
broadly, neither is the subset of the other. For instance, which location gets the request in the first place. Historical-
RUM might be telling you that users are getting great QoE ly, LTM has been essentially blind to user experience; and
from a cluster of virtual servers in Northern Virginia – GTM has been limited to relatively basic local network data
but it doesn’t tell you if those servers are near their capacity (simple “is-it-working” synthetic monitoring for the most
limits, and could do with some help to prevent overloading. part).

Conversely, synthetic data can tell you where the most Application delivery optimization demands not just
abundant resources are to complete a computational, real-time knowledge of what’s happening at both ends, but
storage, or delivery task – but they generally can’t tell you real-time routing decisions that ensure the end user is
whether the experience at the point of consumption will be getting the best experience. Combining LTM and GTM
one of swift execution, or of fluctuating network service makes it simple to
that causes a video to constantly sputter and pause as the
user’s client tries to buffer the next chunk. 1.Improve on Round Robin or Geo-based balancing.
For sure, physical proximity is a leading indicator of superior
Today, though, you can combine the best of both worlds, experience (all else being equal, data that has to travel
as Cedexis has partnered with NGINX and their NGINX + shorter distances will arrive more quickly).
product line to produce a unique application delivery opti-
mization solution. By adding awareness of QoE at the point of consumption,
however, Ops teams can ensure that geographically-boun-
ded congestion or obstructions (say, for instance, peering

OPTIMIZING FOR RESOURCES AND CONSUMERS ALIKE 34


Cloud API

Local load balancer End user


Application-defined
CDN API Data feed load balancing End user

Server data
fusion openmix End user
Application API
Any End user

between a data center and an ISP) can be avoided by the LTM can let the GTM know of its reduced capacity, and
re-routing traffic to a higher-performing, if more geographi- start the process of routing traffic to other alternatives
cally distant, option. before any server instance become overloaded.
In its simplest iteration, the algorithm simply says “so long
as we can get a certain level of quality, choose the closest In essence, here the LTM is telling the GTM not to get too
source, but never use any source that dips below that carried away with QoE – but to check that future expe-
quality floor”. riences have a good chance of mirroring those being
delivered in the present.
2.Re-route around unavailable server instances.
Each data center or cloud may combine a cluster of server 3.Avoid application problems.
instances, balanced by NGINX +. When one of those NGINX+ lets Openmix know the health of the application in
instances becomes unavailable, however (whether through a given node in real-time. So if, for instance, an application
catastrophic collapse, or simply scheduled maintenance), update is made to a subset of application servers, and it

OPTIMIZING FOR RESOURCES AND CONSUMERS ALIKE 35


starts to throw an unusual number of 50xˆerrors, the GTM
can start to route around that instance, and alert DevOps of
an application problem. In this way, app updates can be
distributed to some (but not all) locations throughout the
network, then automatically de-provisioned if they turn out
not be functioning as expected.

Combining the power of real user measurements, hard-


ware health, and application health, will mean expanding
the ability of every team to deliver a high QoE to every
customer.

At no point will user’s requests be sent to servers approa-


ching full use; nor will they be sent to sprightly resources
who can’t actually deliver QoE owing to network conges-
tion that is beyond their control.

It also, of course, will create a new standard: once a critical


mass of providers is managing its application delivery in this
capacity-aware, consumer-responsive, application-tuned
way, a rush will develop for those who have not yet
reached this point to catch up. So take a moment now to
explore how combining the LTM and GTM capabilities of
NGINX+ and Cedexis might make sense for your environ-
ment – and get a step up on your competition.

OPTIMIZING FOR RESOURCES AND CONSUMERS ALIKE 36


Portland, Oregon
421 SW 6th Ave, #700
Portland, OR 97204

Paris, France
15-17 Rue Vulpian
75013 Paris

London, UK
Unit 316, Metal Box Factory
30 Great Guildford Street
London, SE1 0HS

New York
195 Plymouth Street
Brooklyn, NY 11201

San Francisco
995 Market St, Suite 307
San Francisco, CA 94103

Building a faster web. For everyone. By everyone.


Cedexis, now part of the Citrix group, provides web-scale, end-user experience monitoring and real-time traffic routing across multiple clouds and networks.

Cedexis Radar crowdsources billions of real user measurements a day from thousands of popular websites and mobile apps, with traffic routing services based on the
insights this data provides, for the best performance, availability, or cost.

Trusted by over 1,000 global brands including Accor Hotels, Airbus, Bloomberg, Comcast, LinkedIn, Mozilla, Nissan and Rosetta Stone.

www.cedexis.com @Cedexis ©2018 - All rights reserved

You might also like