This ESG Technical Validation documents hands-on testing of Dell EMC Vscale Architecture to verify its ability to reduce operational cost and complexity in today’s data centers. We focus on how Vscale can maximize the value of converged and hyperconverged infrastructure by simplifying deployment and configuration of storage for workloads, easing how Cisco-based networks are extended across multiple sites, and delivering highly-performant hyperconverged infrastructure out of the box.
In an ESG research survey, 25% of IT professionals identified increasing the use of IT infrastructure orchestration and automation tools as a top area for significant investment over the next 12-18 months as they modernize their data centers (see Figure 1). At the same time, 24% of respondents stated that they will also focus their investments on hyperconverged infrastructure (HCI), while 17% will still heavily invest in converged infrastructure (CI).1
Organizations continue to leverage CI and HCI systems to simplify deployment and configuration of server, storage, and networking resources in the data center. However, they still tend to operate these systems as silos. Should insufficient resources exist within a single system to support new workloads, organizations will augment those systems with additional hardware. Unfortunately, this can inadvertently leave unused data center resources from other CI and HCI systems.
Deploying CI and HCI systems as silos also affects how much benefit organizations can gain from automation. Typically, IT staff have automated specific workflows in-house associated with individual CI and HCI systems. Yet, existing automation code cannot simply be applied to similar workflows in other CI and HCI systems. IT would need to apply in-depth knowledge of other systems to properly modify existing automation code, which is time-consuming and can, ironically, detract from any overall savings automation can achieve in the first place. In addition, IT must invest more resources over time to sustain their workflows as versions and technologies evolve, which can drastically increase the cost to automate individual platforms they attempt to improve in the first place.
Addressing these issues requires a solution that can automate workflows for deploying and configuring server, storage, and networking resources to any given workload, regardless of the system in which they currently reside. Any automation enabled should be available out of the box so that its value can be realized quickly with minimal IT effort.
Dell EMC Vscale Architecture
Dell EMC Vscale Architecture is a modular, pre-engineered architecture developed jointly by Cisco and Dell Technologies. It is designed to maximize the utilization of CI and HCI resources deployed within a data center. The architecture pools compute, storage, and networking resources of all physical Dell EMC CI systems so that organizations can deploy, configure, and scale these resources simply and efficiently as workload requirements dictate. This joint effort enables organizations to continue leveraging their Cisco and Dell Technologies deployments, simplifying the integration of Vscale into their existing data centers.
Vscale leverages the following components:
- Dell EMC CI Systems, including Dell EMC VxBlock and its associated compute, storage, and data protection components. (Optional Dell EMC HCI Systems, including VxRail and PowerFlex, can also be supported as part of Vscale.)
- Dell EMC Storage to support applications across the network fabric with required enterprise-level performance and efficiency.
- Dell EMC Data Protection that helps safeguard information from corruption, compromise, or loss.
- Cisco 100G Nexus LAN Fabric that connects multiple Dell EMC systems and modular components to create server, storage, and networking resource pools or cells. It functions as a scalable data center network fabric that supports spine/leaf, edge-core, and storage-area networking (SAN) architectures.
- Cisco MDS for SAN, SAN Insights, and related analytics to support the allocation, deployment, and management of the shared storage pool.
- Cisco Unified Computing System (UCS) Director and Cisco Application Centric Infrastructure (ACI) that enable IT staff to initiate automated workflows developed by Dell Technologies.
- Cisco ACI and Cisco Network Assurance Engine (CNAE) for gaining network visibility and achieving compliance against agility, uptime, and security policy requirements.
- Cisco Tetration to enforce security policies for workloads running on Vscale.
As displayed in Figure 2, organizations have installed Dell EMC CI systems with the intent of deploying and configuring the server, storage, and networking resources racked as a single system to specific workloads. To scale up existing workloads or support new workloads, organizations would then need to purchase additional systems. However, this could lead to underutilized data center resources since they are deployed and managed on a per physical system basis.
Components within Vscale are delivered and supported through Dell Technologies and relevant partners. Dell Technologies and its partners offer additional support through resolution of customer issues, as well as scheduled firmware and software upgrades, pre-tested and pre-validated for interoperability across shared resources.
With Vscale, organizations can now deploy data center resources to any workload regardless of which system those resources reside. Vscale combines server, storage, and networking resources into cells so that organizations can allocate resources as workload demands dictate (see Figure 3). The focus shifts to assigning the amount and type of storage that satisfies workload requirements, negating the need to define the specific CI model or configure the actual physical infrastructure with system-specific workflows and commands.
With Vscale, organizations can efficiently allocate, configure, and reassign the right amount of server, storage, and networking resources in a cloud-like manner. Organizations simply have to manage the amount and type of resources available in each cell.
Similar to how organizations define and deploy virtualized server resources via Cisco UCS Manager, Vscale can configure and allocate storage and networking resources via its policy engine implemented into Cisco UCS Director and Cisco ACI, respectively. Using Cisco UCS Director, organizations create storage virtual arrays and assign their storage resources into storage virtual pools.2 Cisco ACI helps in defining and deploying access profiles for virtualized switch resources defined via switch profiles.
For organizations that deploy workloads into public cloud environments, Vscale also simplifies the extension of an organization’s Cisco ACI-based on-premises network into cloud infrastructure. Vscale simplifies how networking resources can be provisioned and deployed for select workloads that span both on-premises and cloud environments. By instituting common operational workflows, organizations no longer have to manage on-premises and cloud resources with different commands and interfaces, thus increasing operational efficiency. Via Cisco’s ACI Multisite Orchestrator (MSO), traffic policies are applied within IT networks that span both on-premises and cloud environments. Vscale can extend IT environments across multiple public cloud infrastructures simultaneously. Supported public cloud environments include Amazon Web Services (AWS), VMware Cloud (VMC) on AWS, and Microsoft Azure.
ESG Technical Validation
ESG performed evaluation and testing of the Dell EMC Vscale Architecture at Dell Technologies’ facilities at Research Triangle Park in Durham, NC. Testing was focused on three scenarios that demonstrated the simplicity of storage automation, the simplicity of data center resource deployment, and out-of-the-box performance with HCI implementation.
Scenario 1: Simplicity of Storage Automation
Organizations that use Vscale can simplify how they can build and deploy storage topologies spanning multiple Dell EMC CI systems, particularly across multiple Dell EMC VxBlocks. Vscale automates several common storage-related workflows, typically involving moves, adds, and changes. Organizations no longer need to develop in-house automation capabilities that require specific knowledge about the underlying physical storage.
ESG began our testing with the goal of validating how automated storage workflows from Dell Technologies can increase operational efficiency. Out test bed consisted of two “sites” representing geographically dispersed data centers. Each site consisted of multiple Cisco UCS servers along with Dell EMC Unity hard disk drive (HDD) arrays and XtremIO all-flash arrays (AFA). Both sites also contained SANs supported by Cisco MDS 9000 Series switches. Linking the two sites was a Cisco-based IP network comprising Cisco ACI and Cisco Nexus 9000 Series switches. All testing was performed at Site 1. (Please refer to the Appendix – Test Bed Diagram 1 and Table 1 for the test bed diagram and relevant details.) We assumed that storage resources had been configured into storage virtual arrays and pools.
We first navigated to the screen within Cisco UCS Director showing automated workflows enabled by Vscale (see Figure 4). We noted that the options (creating, exporting to and un-exporting from VMware vSphere, removing, and expanding block volumes) do represent the typical operations that a storage administrator might perform daily.
To observe how Dell Technologies enables workflow automation, we clicked on the tile Create Block Volume. We then submitted a service request by entering an existing service request number. Existing service requests represented workflows that have already been submitted into and executed by Vscale. Administrators have the option of creating a new workflow or using an existing workflow with preloaded inputs. ESG noted how this can help an administrator save time in performing a storage operation by leveraging what has already been executed.
ESG then proceeded to examine the preloaded inputs and determine whether any needed to be changed for our current task. Key values to review included Datacenter, Virtual Pool, and Virtual Array (within the chosen Datacenter), Volume Size, and Number of Volumes (see Figure 4.). We noted that Vscale presented the Dell EMC storage array models underlying both the virtual pool and virtual array within the chosen data center. We inputted the values to create a single 10GB storage volume. Once we verified and submitted the service request, ESG verified that the workflow executed automatically and completed in a matter of minutes.
ESG also performed similar steps for other workflows, such as exporting and expanding a storage volume. We executed similar steps from our previous test and observed that those workflows also completed automatically in minutes.
ESG clearly saw how Vscale decreases the time and effort for an administrator to complete storage-related tasks. Even if an organization has automated workflows in-house, the amount of time required to learn specific storage array models and script the appropriate code requires time and effort. Multiple people may be required to contribute to a single script, as they all possess knowledge about the required tasks. If an administrator wanted to automate a similar workflow for another storage array model, that would require modifying that script to accommodate different given commands and interfaces for operating and managing that array. Finally, that level of automation is limited to the rack in which the storage array is located. The time and effort spent to automate a workflow in-house may not be worth the incremental benefit realized.
Why This Matters
Organizations that have deployed CI systems in their data centers have benefited by easing the deployment and configuration of data center resources for a given workload. However, as more CI systems are deployed, optimizing the deployment, allocation, and management of these resources becomes cumbersome as these systems are regarded as silos, each with their distinct workflows. This mode of operation is not ideal for organizations that wish to deploy and configure these resources in a flexible, scalable, and consistent manner.
ESG validated that Vscale enables automated workflows for common storage operations across virtualized storage arrays and pools. The level of automation achieved supports organizations in deploying, configuring, and allocating storage resources easily and applying that automation repeatedly (when required) regardless of the underlying physical storage array models. ESG clearly saw how Vscale can decrease the time and effort an administrator spends on completing storage-related tasks.
Scenario 2: Simplicity of Extending Networks Across On-premises and Multiple Cloud Environments
Extending IT networks that span between multiple sites, either on-premises or in the public cloud, requires that organizations coordinate tasks and workflows unique to each environment. However, this leads to a lack of operational efficiency and consistency. With Vscale, organizations can execute networking-related workflows simply and repeatedly, regardless of whether the underlying infrastructure is based on-premises or within a public cloud environment.
ESG tested how Vscale simplifies how organizations deploy IT networks spanning on-premises data centers and the cloud. We used the same test bed described in Test Bed Diagram 1 and Table 1 located in the Appendix. Testing utilized both sites.
We began by accessing UCS Director to navigate to Cisco MSO (see figure to the left). We observed that Vscale supplies a library of networking-related automated workflows available out of the box. Again, these are common workflows (moves, adds, and changes of networking resources) typically performed in daily IT operations.
To examine the workflow steps in flowchart form, ESG chose the workflow named “ACI MSC – Create Bridge Domain L3 and EPG – v4.” ESG noted immediately how Vscale can simplify workflow execution via automation, as we noted the number of steps to be completed, the execution loops, and decision trees involved in completing this workflow across multiple systems. Should this workflow be executed manually, an administrator must consider the number of management interfaces to be used and the amount of in-depth knowledge of the switches to be acquired.
ESG proceeded to execute the workflow for creating a bridge domain3 between two Cisco ACI-enabled sites (see Figure 5). We followed a similar process from our previous testing and chose an existing Service Request associated with a workflow for creating a multi-site network. Each site is defined by its Cisco API Controller (APIC).4 We also selected values defining the virtual routing and forwarding (VRF) instance, the VLAN, and the subnet. Once we had input the required parameters, we submitted the request.
After executing the workflow, ESG navigated to the Cisco MSO and verified that the two chosen sites were connected via an L2 bridge domain (see Figure 6). Time to complete the workflow was only a few minutes.
ESG noted that it took significantly less time for the workflow to execute than the typical process of executing tasks via multiple interfaces, each designed with inherently different workflows and navigation patterns. Obtaining approvals for any network change, coordinating efforts within the IT team, then accessing the relative management interfaces to execute discrete steps (and repeating these steps in several cases) can easily consume weeks of time and effort.
Why This Matters
Extending existing Cisco ACI-based networks to encompass multiple sites, either on-premises or in the cloud, is not an easy feat. The problem lies in the differences between the operational and management models for deploying and configuring resources from the different underlying architectures and technology.
ESG validated that Vscale enables automated workflows for common Cisco ACI-based network operations across multiple data centers and cloud environments. The time and effort required to gather and apply knowledge of the underlying technology and interact with differing management interfaces, especially when extending networks across multiple public clouds, is significantly minimized. Leveraging Vscale’s workflows enables organizations to achieve efficiency and consistency in their IT operations.
Scenario 3: Out-of-the-box Performance for HCI Implementation
When organizations use software-defined storage (SDS), networking performance becomes a requirement to ensure that performance does not degrade. Misconfiguration of the network can cause higher latency, performance issues, and data unavailability or loss. Dell EMC PowerFlex mitigates these issues via automating workflows to deploy and configure HCI nodes, eliminating any chance of network issues arising due to manual error.
ESG tested how Dell EMC PowerFlex’s automated workflows for configuring SDS prevent issues that arise if HCI nodes were configured manually. The goal was to assess the performance impact considering manual errors.
Located in Research Triangle Park, the test bed was a standard data center Layer-2 (L2) Ethernet virtual private network (EVPN) consisting of two Dell EMC PowerFlex systems. Each system contained eight Dell EMC PowerEdge R640 server nodes and a Cisco Nexus N93180YC top-of-rack (ToR) switch. All server nodes were connected redundantly to each ToR switch via 25 Gb/s links. The network was deployed in a spine/leaf architecture consisting of Cisco 9000 Series switches. (Additional test bed details and diagram are in the Appendix.)
For our tests, we leveraged four VMware virtual machines (VMs), each loaded on four separate nodes; the remaining nodes remained inactive. Traffic on all VMs was generated using FIO, an I/O tool used for benchmark and stress/hardware verification. Test traffic contained a random mix of reads and writes and various block sizes to mimic a variety of supported applications.
ESG first tested the performance impact of maximum transmission unit (MTU) configuration mismatch on storage performance. If the allowed MTU size of the L2 traffic packets is too big to be transmitted on a given network connection, packet loss will most likely occur. Subsequently, SDS performance will degrade. We began by generating traffic until the I/O workload reached a steady state of 1.5GB/s, as observed on the PowerFlex Manager interface (see Figure 7). All four pairs of data paths were running, and no storage rebuilds were recorded.
On the switch side, ESG then changed the MTU of one data path from 9,126 to 1,500 bytes. SDS performance began to degrade; eventually, the I/O workload decreased to 654.2 KB/s. Storage rebuilds consumed 10.4 MB/s of the existing workload, contributing further to the overall performance decrease. One data path also went down, decreasing the number of paths on which traffic could travel between the switch and storage (see Figure 7).
ESG then tested how manual port misconfiguration can negatively affect SDS performance. Figure 8 shows that performance was initially measured at 2.6 GB/s. We set up the test bed so that port speed matched the ToR switch port speed. We then manually changed the port speed from 25 Gb/s to 10 Gb/s.
We observed that SDS performance again decreased to as low as 40.2 MB/s in a matter of seconds. Once that error was recognized by the management system, the HCI node clusters redistributed the load so that performance recovered to its initial levels. However, the risk remained that traffic was dropped during that short time interval and could not be recovered. Alternatively, avoiding this manual error by using PowerFlex Manager automation workflows can reduce this risk.
Why This Matters
When leveraging SDS on HCI nodes, correct network configuration is crucial to maintain high performance.
ESG validated that Dell EMC PowerFlex, as part of the Vscale Architecture, helps organizations to maximize the use of their HCI storage resources. By automating workflows to eliminate manual errors in configuring networking resources, organizations can reduce the risk of degraded storage performance when supporting multiple critical business applications.
Highmark Health, headquartered in Pittsburgh, PA, is the parent company of Highmark Inc., a hospital system, and other businesses. Highmark Inc. and its subsidiaries and affiliates provide health insurance to more than 5.6 million members in Pennsylvania, West Virginia, and Delaware as well as dental insurance, and related health products through organizations such as Blue Cross and Blue Shield. Serving up to 60,000 users at any time, Highmark Health’s IT environment currently consists of two geographically dispersed data centers networked via a Cisco ACI fabric.
Highmark Health had been evaluating new HCI platforms for supporting their virtual desktop infrastructure (VDI) environment. However, the IT team had to consider how easily the new equipment could integrate into the existing Cisco environment. Specifically, integrating a new HCI platform into the existing Cisco environment would consume time and effort, ultimately degrading the IT’s team ability to meet time-to-market demands (i.e., deliver virtual desktops without unnecessary delay).
Highmark Health decided to install Dell EMC VxRail HCI nodes and leverage the Vscale Architecture. They have begun with a 200 physical node deployment across both data centers.
Highmark Health could easily integrate the Dell EMC VxRail nodes into the existing Cisco ACI fabric. Otherwise, the networking team would need to individually configure the switches on each VxRail node and connect them to the Cisco ACI fabric via manual processes. Vscale automates configuration workflows to integrate the VxRail nodes into the existing Cisco environment. Performance issues are greatly minimized given the lack of misconfiguration errors. The automation also helps to bring up the Dell EMC VxRail nodes quickly, helping Highmark Health to meet time-to-market demands for delivering virtual desktops to any end-user with minimal delay.
The Bigger Truth
Organizations continue to deploy both CI and HCI systems to simplify the deployment, configuration, and management of data center resources. These systems are typically pre-engineered and pre-configured so that they support a predefined set of workloads. However, as more CI and HCI systems proliferate in the data center, organizations still face the challenge of efficiently allocating and managing their resources, without necessarily having to purchase additional racks of systems. It would be ideal for organizations to deploy, allocate, and configure server, storage, and networking resources, efficiently and consistently, regardless of which physical system they reside in.
The Dell EMC Vscale Architecture, leveraging a combination of Cisco and Dell Technologies’ solutions, enables organizations to build virtual pools of data center resources and assign them to workloads in a flexible and scalable manner. With automated workflows, organizations can deploy, configure, and maximize the use of server, storage, and networking resources within Dell EMC CI systems, regardless of where those resources are racked physically. This architecture is pre-engineered and pre-tested prior to delivery to help organizations minimize both time to deploy and time to value.
ESG validated that Vscale can help organizations to maximize the value extracted out of their Dell EMC CI and HCI systems by automating common storage allocation and configuration operations without the need for in-house development, simplifying how Cisco-based networks can extend across multiple on-premises data centers and cloud environments, and ensuring out-of-the-box performance when implementing multiple Dell EMC HCI nodes for supporting multiple applications. All tests revealed that Vscale offers organizations the opportunity to introduce both simplicity and consistency to daily IT operations, subsequently maximizing performance for any workload.
While ESG tested with Vscale as a part of Cisco UCS Director, readers should keep in mind that Cisco is moving UCS Director functionality onto Cisco Intersight. For organizations to reap the full benefits of Vscale, they may want to evaluate how the transition will occur to avoid any feature or functionality gaps.
The business value that ESG has validated that the Dell EMC Vscale Architecture can deliver—less time and fewer resources spent on allocating and configuring data center resources for any workload while increasing value extracted out of deployed CI and HCI systems—is clear. For organizations that see automation as a way to help reap these benefits while leveraging both Cisco and Dell Technologies deployments, the Dell EMC Vscale Architecture is worth closer evaluation.
1. Source: ESG Master Survey Results, 2020 Technology Spending Intentions Survey, January 2020.↩
2. A storage virtual array defines logical storage arrays characterized by port groups or performance groups that span multiple physical storage arrays. A storage virtual pool defines a resource pool with common storage capabilities such as IOPS, throughput, and compression/deduplication rates, and replication strategy.↩
3. A bridge domain is a set of logical ports in a Cisco ACI network that share the same flooding or broadcast characteristics. Like a virtual LAN (VLAN), bridge domains span multiple devices.↩
4. A Cisco API Controller (APIC) is a single point of policy and management of a Cisco ACI switch fabric.↩
ESG Technical Validations
The goal of ESG Technical Validations is to educate IT professionals about information technology solutions for companies of all types and sizes. ESG Technical Validations are not meant to replace the evaluation process that should be conducted before making purchasing decisions, but rather to provide insight into these emerging technologies. Our objectives are to explore some of the more valuable features and functions of IT solutions, show how they can be used to solve real customer problems, and identify any areas needing improvement. The ESG Validation Team’s expert third-party perspective is based on our own hands-on testing as well as on interviews with customers who use these products in production environments.