Professional Documents
Culture Documents
X
APPLIANCE AND RACK
DESIGN STUDENT
GUIDE
DOWNLOADABLE CONTENT
DOWNLOADABLE CONTENT
PowerFlex 4.X Appliance and Rack Design Student Guide
Resources 84
PowerFlex 4.x Appliance and Rack Design Resources 84
Glossary 86
The Dell Professional Services team is made up of both Dell internal and
partner services personnel. The core plan and design team consist of
project managers and solution architects for the design effort. The solution
architect works closely with the pre-sales team and the PowerFlex
implementation team for deployment. The core team of project managers
and solution architects act as a bridge between Dell pre-sales and the
PowerFlex solution deployment.
SDS
SDC
The Storage Data Client is a block device driver that exposes shared
block volumes from the SDS to the operating system. The SDC runs on
SDT
The Storage Data Target manages host connections and controllers that
are connected over NVMe/TCP. SDT sends administrative and I/O
commands to the SDS programmatically such that the SDS is oblivious to
the source of the I/O. This makes traffic from an SDT look like it came
from an SDC.
The web version of this content contains a video.
MDM
The Meta Data Manager is the authority that controls and tracks data
storage ownership, mapping, and protection. As volumes are created, the
MDM provides the information application servers need to connect to the
cluster’s virtualized storage.
The web version of this content contains a video.
FSN
Data Layout: Meta Data Manager map applies granularity rules and
Protection Domain rules to evenly distribute blocks of
• Medium
data across every node.
Granularity
Medium Granularity is 1 MB blocks geared toward
• Fine speed and performance while Fine Granularity is 4 KB
Granularity blocks geared toward compression and snapshots.
File Support File Controllers are dedicated physical nodes that host
NAS servers.
Review the PowerFlex 4.0 course to learn more about the PowerFlex
features and functions in detail.
• PowerFlex 4.0 Administration
on VMware virtual storage area network (vSAN) while the PFMC 2.0 is
based solely on PowerFlex with Kubernetes. PFMC 2.0 can use either
a single management node, with raid, or use 3 plus management
nodes with a Powerflex cluster. PFMP also runs on software-only
PowerFlex solutions where customers use their own management VMs
or Co-Resident storage nodes.
Review the PowerFlex 4.0 course to learn more about the PFMP and
PFMC in detail.
• PowerFlex 4.0 Implementation
Data Gathering
This Oracle RAC on PowerFlex rack white paper outlines how customers
can deploy Oracle RAC databases on PowerFlex rack to meet
performance, resiliency, scale, and availability requirements.
The customer data center virtual storage and compute needs vary. One
customer may require storage-only custom nodes to meet their needs.
Another customer may require a full integrated rack solution because of a
different need.
PowerFlex rack
PowerFlex appliance
PowerFlex appliance has a smaller starting point than rack, and still scales
to hundreds of nodes. PowerFlex appliance allows customers to use a
broad set of supported networking options and can be added to existing
networking infrastructures. However, PowerFlex appliance is limited to
supported networking configurations.
As with the nodes, the options for deploying the models vary as well.
Based on which nodes make up the model, there can be three different
deployments: Two-Layer, Hyperconverged, and Mixed.
Two-Layer
The SDC requests data from the mapped volumes of the storage pool.
HCI
In the HCI configuration, both the SDC and the SDS are installed on the
same host. HCI deployments maximize hardware utilization and reduce
infrastructure requirements.
In the HCI deployment model, the node SDS contributes storage to the
storage pool. The same node SDC makes requests for data from the
mapped storage pool volumes.
Mixed
The HCI nodes contribute both compute and storage resources from the
same host. The singular SDC nodes provide compute resources, making
request for data from the storage pool volumes. The singular SDS nodes
provide storage resources, contributing storage data to the storage pool.
Rack Two-Layer
Appliance HCI
A small retail chain wants to update their IT infrastructure. The retail chain
does not need a lot of compute and storage capabilities. However, the
retail chain does require handling a spike in transactions at peak times
during the year. They have a limited budget and space available for
hardware. Other storage solutions do not meet their performance needs.
The retail chain provides their own networking and management
environment.
Scenario
A local hospital must process and store many patient records. The
hospital needs fast, high-performance storage for hundreds of medical
records for personnel to access quickly. The hospital must also expand
their storage quickly as their patient records increase. The hospital is
looking for a Dell system that is a fully engineered solution as it includes
supported networking devices.
Scenario
PowerFlex uses TCP/IP and NVMe/TCP protocols to offer block and file
services.
The differences on how the PowerFlex custom node, appliance, and rack
handle networking are:
FNA is the Dell preferred network option for PowerFlex appliance since it
is a fully tested system. For the PowerFlex appliance FNA, the customer
can:
Aggregation/Access
Spine/Leaf
Review the VLAN and Data Transport Protocol Specific Details to learn
more.
1The recommended maximum subnet size for each rack network is under
27 subnets. The recommended maximum subnet size for each row
network is under 21 subnets.
Review the table for the supported node types and switch topologies for
multi-subnet and multi-VLAN PowerFlex 4.5 rack or appliance
implementations.
The VLAN requirement for the PFMC 2.0 management nodes is changed
in PowerFlex 4.5 because of multi-subnet. PFMC 2.0 is reliant on a series
of management virtual machines (VMs) to control the PowerFlex cluster
and provide the block services for each node. A new VLAN subnet, named
pfmc-mgmt-130 (VLAN 130), is used in PowerFlex 4.5 for all
management VMs when implementing multi-subnet.
• VMware VCSA
• Jump Server
• PowerFlex Manager (PFxM)
• Dell CloudLink
• Secured Connected Gateway (SCG)
• Policy Manager
• PFMP (Kubernetes Cluster)
For all the other node types, the use of VLAN 105 for multi-subnet is
allowed. However, the VLAN 105 implementation will cause the
management VMs in PFMC nodes to break after vMotion migration. The
use of the new VLAN 130 remediates the ESXi Management and
management VMs issue for a PFMC node multi-subnet implementation.
ACI
NSX-T Ready
If the edge nodes and controller are part of the appliance, then the
required hardware components are installed at the Dell factory. The NSX-
T software for the controller and edge nodes is installed in the Dell factory
or onsite by the Dell Implementation team.
NSX-T edge nodes can run as virtual appliances or bare metal nodes.
However, PowerFlex rack only supports NSX-T edge nodes as virtual
appliances. The edge nodes connect directly to the aggregation switches.
Inputs include:
1. Minimum Medium Granularity (MG) Capacity
2. By default, the results are sorted by relative price. To sort by another
metric, click to select space, speed, and so on.
3. Network Speed: 25 Gbps or 100 Gbps
4. Workload Read/Write Percentage
Filters include:
1. Storage Data Layout: Medium Granularity or Fine Granularity
Sizer Results
Sizer considers the inputs and filter and shows the results.
1. Click one of the results in the right panel to show its details in the left
panel.
2. View the details from multiple viewpoints, including raw and usable
storage, workload, compute, network, and availability.
3. In the example below, the Availability option shows the amount of time
it would take to rebuild a failed node and failed drive.
4. Export a PDF to include all or selected information about the
configuration.
Advanced Sizer
To use the Advanced Sizer, clear the Simple Sizer toggle. The advanced
mode includes additional filters and inputs. Options are available to select
replication, specific PowerFlex products, host connectivity, chassis model,
and specific CPUs, among others.
Two concepts to consider when using the Sizer to design the PowerFlex
solution are storage node reliability and availability.
Follow the rules and best practices to determine the size and scale of the
best PowerFlex solution for the customer.
After initial requirements are gathered from the customer, there will be
more conversations to find the design that best fits their needs. View the
video below to see one possible design to the Activity 1 scenario.
View the video below to see one possible design to the Activity 2 scenario.
See the Enterprise Project Services Concepts and Features course for a
one-hour overview video of the EPS tool.
Manage a Project
After the PowerFlex solution has been sold, the project moves to the
implementation phase in EPS. Use EPS to design and manage the
project.
Manage a Project
Within an EPS project, there are one or more customer orders. EPS
receives customer orders from Financial Force, or orders can be created
manually.
During the Design phase, Dell team members, partners, and customers
collaborate to complete the design of the PowerFlex solution. In EPS, this
process follows one of the PowerFlex templates: PowerFlex appliance 4.0
or PowerFlex custom node 4.0. The information entered into the template
is used to correctly implement the system on site.
The pre-deployment checks are a part of the overall design process that
provides the status of various tasks that are associated with the
deployment.
The report indicates the status of the deployment site, including data
center readiness, network readiness, and other relevant factors.
1. Create a pre-deployment checks report, name it, and then publish the
report.
2. Complete the required pre-deployment checks tasks.
3. Submit a final pre-deployment checks report after working with the
customer to refine the report.
The EPS project team member completing the design of the PowerFlex
solution creates the pre-deployment checks report. After the team member
creates the report, then they go to publish the report. Publishing the report
allows the customer to participate by providing the network check details.
Step 2: Name the report and then select the Create button.
Select the Manual network & site checks tab to proceed with the
publishing process.
Publish button is selected on the Manual Networks & Site Checks page
in order to proceed with the publishing process.
Publishing a report does not mean that the report is submitted for final
approval. Publishing means that the report is made available in EPS for
data entry and editing by all team members including the customer.
The left navigation menu and the data field do not provide any data entry
capabilities, indicating a lack of input options. Until the report is published,
details about the network and site cannot be entered or captured.
Upon selecting Yes, publish in the Publish forms for manual checks
dialogue box, the report is published, enabling the technician and
customer to proceed with further tasks.
Once the report is published, all the forms in the Manual network & site
checks page are active. Each section of the page can be populated with
data from previous work, such as design (Project Details).
In addition to the Manual network & site checks page, the Automated
network checks page also becomes active. The customer performs the
automated check by accessing the page to complete the network check
onsite. If the customer decides to skip the automated network check, they
must provide a valid reason in EPS.
The check that is completed in the automated page gets populated in the
section that is named Network check in the final report.
Once the automated network check is done, the card marks the
Automated network checks section as Completed.
• Infrastructure
• Data center switches
• Power and cooling
• Cabling and transceivers
• Data center access
During the Manual network & site checks task, the details that are
entered are in draft mode. The left panel tracks data entry progress,
showing check marks for completed details and indicating the remaining
information to be entered.
The Save button is selected to maintain the integrity of the data and allow
for interim data entry.
To change the details on a report that has not been submitted, select the
Modify option. At any time, the data in this section can be discarded by
selecting the Discard changes button and reentered if the report is in
draft mode.
A technician exports the report during the data entry effort in the Manual
network & site checks task. The report export is used when the
customer cannot work with the published EPS report.
The Manual network & sites report is downloaded as a ZIP file. The
exported file is shared with the customer to view offline.
At some point in the Manual network & site check data entry, the
customer must provide some revisions or additional data. The Pre-
deployment checks card displays a Needs input status when the
customer must provide the details.
After the customer completes their review and revision of the exported
report, they send the file back to the team member technician. The
technician imports the revised customer file in the Manual networks &
site checks section.
The size of the imported ZIP file must be less than 10 MB. The imported
file must be the same exported file from the offline bundle of the same
project.
After verifying and accepting the design, the customer hands off the
project to the PowerFlex Implementation team. The team then deploys the
PowerFlex solution at the customer site.
Once all the tasks in EPS are completed, the project manager closes the
project.
The EPS for Deployment Partners Tool, also known as EPS offline, is a
version of the EPS online tool with limited capability. The focus of the EPS
offline is to allow Dell partners to conduct PowerFlex custom node or
appliance design captures on their own without support from Dell
Professional Services. The Dell partner solution architect captures the
design details with the customer and provides the information to
deployment partner engineers.
The Customer Configuration Center (C3) online design tool provides the
PowerFlex rack design survey page. A PowerFlex rack SA or a Dell
partner TA and customer work together to capture a PowerFlex rack
design. The output from the C3 online tool is a Logical Configuration
Survey (LCS).
The LCS defines the configuration of the PowerFlex rack system for a
deployment engineer. The LCS is the blueprint that takes account of:
Project Details
The PowerFlex rack details are provided in the LCS through categories of
information. Each category is collapsible for ease of use.
An SA or TA use the Opportunity ID and the Order ID to look up the
details about the configuration of the PowerFlex rack system in C3.
System Information
Deployment engineers can view the details of the deployment
requirements in the System Information category.
Important details such as RCM version, type of PowerFlex rack
implementation, network architecture, network devices, and number of
servers that are required are provided. These details are captured in the
C3 online tool by an SA or TA and customer.
Storage Layout
The Storage Layout information provides the deployment engineer a view
into the required PowerFlex rack cluster storage implementation.
Information on the Protection Domain, Storage Pool, and Nodes (storage)
plus drive types is captured. In this category, the PowerFlex Protection
Domains for the SDS nodes that are being modified or added to the
cluster are identified. SDC nodes are not applicable to this category.
System VLAN
The System VLAN category provides all the VLAN information that is
captured during the PowerFlex rack design process. IP addresses for the
System Management, Hypervisor Management, vMotion, PowerFlex Data
#1, and PowerFlex Data #2 VLANs are also identified.
Fast Track
The Fast Track category relates to the auto-population of IP addresses for
VLANs. Deployment engineers can edit any IP address that is auto-
populated.
Device Information
The LCS provides the network switch information in the Device
Information category. An engineer can look up the device type, IP
address, and provide a hostname for the switches that are used in the
PowerFlex rack network topology.
Cluster Configuration
Deployment engineers can look up the data center and cluster
identification information in the Cluster Configuration category of the LCS.
For PowerFlex rack expansions, the engineer ensures that the added
nodes are configured to meet the cluster configuration requirements.
Core Environment
The Core Environment category of the LCS provides deployment
engineers the customer domain, DNS, and NTP server information. The
Domain name and IP address are identified, along with DNS servers
hostname and IP address and NTP servers hostname and IP address.
Survey Finalization
In the Survey Finalization category of the LCS, the deployment engineer
can add notes as needed. Also, the engineer can review all the categories
with the customer. When changes are needed, the engineer obtains a
sign-off that the changes meet with the customer approval.
Resources
The following list contains links to the resources for this course:
LACP allows the switch to use either Cisco Virtual Port Channel (VPC) or
Dell Virtual Link Trunking (VLT). LACP provides bundling of multiple port
channel connections to transfer a large pipeline of data at one time. The
connection appears as one port channel link to the receiving device,
whether it is a server or another switch.
Co-residency
The PowerFlex storage-only nodes in a protection domain runs the
management of the system.
Fault Set
A Fault Set is a logical entity that contains a group of SDSs within a
Protection Domain (PD) that have a higher chance of failing together. By
grouping the SDSs into a Fault Set, PowerFlex mirrors the data for a Fault
Set on separate SDSs that are outside the Fault Set. In this way,
availability is assured even if all the servers within one Fault Set fail
simultaneously.
Fault Set
A Fault Set is a logical entity that contains a group of SDSs within a
Protection Domain (PD) that have a higher chance of failing together. By
grouping the SDSs into a Fault Set, PowerFlex mirrors the data for a Fault
Set on separate SDSs that are outside the Fault Set. In this way,
availability is assured even if all the servers within one Fault Set fail
simultaneously.
IOPS
IOPS (input/output operations per second) is the standard unit of
measurement for the maximum number of reads and writes.
Mixed Deployment
Hybrid hyperconverged deployment consists of hyperconverged, compute-
only, and storage-only nodes. Some nodes contribute both compute
resources and storage resources (hyperconverged nodes), some
contribute only compute resources (compute-only nodes), and some
contribute only storage resources (storage-only nodes).
Nodes
Nodes are the basic hardware units that are used to install and run a
hypervisor and PowerFlex software.
Nodes
Nodes are the basic hardware units that are used to install and run a
hypervisor and PowerFlex software.
PowerEdge R650
The PowerEdge R650 platform is a 15G single rack unit. As such,
PowerEdge R650 provides great CPU core density for usage in a
compute-only, or hyperconverged node use-case. PowerEdge R650
provides ten drive slots for an overall maximum capacity of 76.8 TB. While
more expensive than two rack unit systems, two R650 systems can
provide a similar storage density to the R750 and R850.
PowerFlex
PowerFlex is an enterprise-class, software-defined block and file storage
solution that is deployed, managed, and supported as a single system.
PowerFlex
PowerFlex is an enterprise-class, software-defined block and file storage
solution that is deployed, managed, and supported as a single system.
PowerFlex appliance
PowerFlex appliance has a smaller starting point than PowerFlex rack, but
scales to hundreds of nodes. PowerFlex appliance allows customers to
use a broad set of supported networking options and can be added to
existing networking infrastructures. However, PowerFlex appliance is
limited to supported networking configurations. PowerFlex appliance
comes with licensing for a PowerFlex system and a unified management
platform.
PowerFlex Cluster
A collection of multiple nodes, either SDS, SDC, or HCI, that
communicates with each other to perform set of operations.
PowerFlex Device
Local, direct attached block storage (DAS) in a node that is managed by
an SDS and is contributed to a storage pool.
PowerFlex Manager
PowerFlex Manager automates deploying, configuring, managing, and
upgrading a PowerFlex system. The PowerFlex Manager Dashboard
provides an overview of the PowerFlex system. PowerFlex Manager
allows administrators to quickly view the status of the system hardware
and software components.
PowerFlex rack
A fully engineered rack-scale system with integrated networking.
PowerFlex rack has a larger starting point, and scales to hundreds of
nodes and racks. PowerFlex rack comes with licensing for a PowerFlex
system and a unified management platform.
PowerFlex System
A PowerFlex system is the collection of entities managed by the Metadata
Management (MDM) cluster.
Protection Domain
A Protection Domain (PD) is a group of nodes or SDSs that provide data
isolation, security, and performance benefits. A node can only participate
in one PD. Separate PDs can be created for different node types with
unequal performance profiles.
Protection Domain
A Protection Domain (PD) is a group of nodes or SDSs that provide data
isolation, security, and performance benefits. A node can only participate
in one PD. Separate PDs can be created for different node types with
unequal performance profiles.
Storage Pools
The Storage Pools are a subset of physical storage devices within a
Protection Domain (PD). Each storage device belongs to a Storage Pool.
Dell Technologies recommends having the same type of storage devices
within a Storage Pool to ensure that the volumes are distributed across
the same type of storage within the PD.
Storage Pools
The Storage Pools are a subset of physical storage devices within a
Protection Domain (PD). Each storage device belongs to a Storage Pool.
Dell Technologies recommends having the same type of storage devices
within a Storage Pool to ensure that the volumes are distributed across
the same type of storage within the PD.
SupportAssist
SupportAssist is a proactive monitoring software with automatic failure
detection and notifications for Dell PCs, tablets, and servers.
SupportAssist is free of charge, secure, and streamlines traditional
support routines.
Two-Layer Deployment
In a two-layer deployment, the SDS is installed on a separate host from
the SDC. The front-end (client) is separated from the back-end (storage)
data traffic.Two-layer deployments allow compute and storage resources
to grow independently. PowerFlex compute-only nodes host end-user
applications. PowerFlex storage-only nodes contribute storage to the
system pool.
VCSA
Volume
Analogous to a LUN, a volume is a subset of a storage pool’s capacity
presented by an SDC as a local block device. A volume’s data is evenly
distributed across all disks comprising a storage pool, according to the
data layout selected for that storage pool.