Build vs. Buy? Field Insights on Speed, Deployment, Scalability, and Savings

Table of Contents

A Pragmatic Approach to Change and Data Center Modernization

IT teams in government agencies have a lot of priorities to juggle. Federal initiatives propel them toward greater IT efficiency and data center modernization. They also encourage embracing multi-cloud forms of IT service delivery, where it makes sense. And, of course, IT organizations must do all this while still being called upon to support ever more challenging federal workloads—from demanding virtual desktop infrastructure (VDI) deployments to tactical, field-based operations and the still-present needs of remote offices.

The pressure is on to make progress across the board, often under still-limited IT budgets.
How best to move forward? Increasingly, more agencies are adopting a pragmatic approach to modernization that also helps them lay a foundation for hybrid cloud. The preference should be “mission first” and that is the conundrum within any IT shop. Because the industry has continued to grow faster with more options, customers now have millions of options, many of which have little if any to do with your base mission.

When making choices the Federal Government has been extremely diligent in creating lists of hardware and software and services that are approved for use. The issues only come after that list, “in what version, configuration and use case can you honestly hope that these pieces, parts, and solutions work together”.

The road to such savings is not always clear. While ‘the road less travelled’ allows for innovation and creativity, it also tends to come with its own pit falls. Creating your own solution that will no longer be supported by the manufacturer of the hardware, software or solution provider or eventually will be outgrown by the mission. This is just one of the findings by IMPRES Technology Solutions in our implementations with federal agencies.

Unfortunately, we see many agencies struggle to implement the right underlying hardware, software and solution components to best support their long term goals.
The challenge often seems to center around a common dilemma:

• Should our agency “grow its own” underlying, converged infrastructure with its own mix of hardware components (commodity servers/storage, I/O controllers, CPU/memory, NICs, etc.)?
• or
• Should we use a turnkey product with the parts pre-integrated, pre-configured, then tested and validated before shipping?

To make it worse, now there are multiple reference architectures, ready nodes, niche storage devices, appliances and multiple versions of building blocks by every manufacturer. How do you know what does what and when do you need to select one vs another?

Back in the day, people used to build their own desktops, we later found it is generally much more cost effective to purchase such base units as turnkey and delivered. Just like building a car, most people don’t buy their own engine, drivetrain, transmission and fuel injection/carburetor etc. and attempt to build their own anymore. The mission is the goal, not the commodity parts that help us get there.

To help address this dilemma, this paper shares IMPRES’ field experience with a few vSAN federal customer environments. It also draws on other user experiences and government use cases with vSAN.

Figure 1 Build vs Buy Magic Decoder Ring

From low to high integration and Open to Highly Integrated hypervisor. If you ever think you want to build your own airplane just to fly across the country vs just getting on a plane, then the Reference Architectures are for you.

Reference Architectures

provide all the versions of software, hardware, solutions that are on the Hardware Compatibility List and that each vendor has gone to some length to verify it will “work” with the solution you are looking for. Much like saying that Windows only needed 32Mb of RAM to run, that only worked if you didn’t need anything more than NOTEPAD on your system. While the individual components are tested, and the theory behind how they go together is generally rather sound, this can become a major issue if you want to deviate or live on the edge of mainstream with your solutions. Versions of firmware or a simple bios change can cause your specific application not to perform at all on the hardware you chose off the vendor’s Hardware Compatibility List (HCL) and Reference Architecture.

One of the benefits of vSAN technology is its ability to simplify and centralize the management of distributed, underlying storage into a virtual pool or cluster. Part of the appeal of this design is that vSAN technology may even allow organizations to incorporate some of their legacy hardware already in use within their federal data center environments.

This leads some agencies to conclude that buying their own, underlying commodity parts may be a more economical way to go when building out vSAN infrastructures. This means, for instance, the acquisition and implementation of their own commodity servers, NICs, CPUs, etc.

For those who choose this type of “DIY” scenario, VMware offers a several-hundred-page vSAN Hardware Compatibility List (HCL). It also offers a large body of documentation, best practices, sizing and configuration tools and architectural guidance for such scenarios.

The truth is, however, that we often see organizations running into challenges when they attempt to DIY their own. Such challenges can include:

• Unexpected costs
• Unanticipated configuration issues
• Significant time required to test and troubleshoot issues
• Unplanned complexity for less-experienced operations team members
• Multi-month deployment delays
• Sizing or read/write performance issues for heavy and mixed workloads
• Unexpected network issues that may slow expected vSAN performance

Dell EMC Ready Nodes

VMware, like many other of the vendors in the community, will “strongly recommend” using something that is pre-validated, pre-certified so that the chances of your compatibility are much greater. The Dell EMC Ready Nodes are designed to provide a predictable outcome and scalability of components.

The big win for the Ready Node (such as with the vSAN), is that of end user performance and savings. This was the case in the Principled Technologies report, “Support more database users while minimizing datacenter sprawl,” August 2017, which compared Dell EMC software-defined storage solution with 8-node configurations to a similarly configured flash storage-based HPE solution of DL360s server nodes and 3PAR 8450 storage using the DVD Store 2 database workload benchmark test.

The key reason for leveraging these Ready Nodes is that actual performance will vary based on configuration, usage and manufacturing variability. It also was shown that customers handle up to 1.8X more database orders per minute. Customer using Ready Nodes noted 87% increase in their speed of business much because the need for “making it work” was removed from their realm of worry.

As shown in Figure 1 above, the vSAN Ready Node contains an appliance hardware from Dell EMC, but even better for government procurement cycles, the solutions come with a single SKU and BOM for easy ordering while still offering sufficient flexibility to “tweak” underlying elements to achieve custom results.

The customer chooses the Hard drives, Memory and Network capabilities for the Ready Node. There are various performance/capacity mixtures to support the use of solid-state disks (SSD), hard disks drives (HDD) and/or NVMe disks.

While other vendors also offer VMware vSAN Ready Nodes, the solutions we are discussing in this paper are jointly validated and certified by VMware and Dell EMC, involving:

• More than 5 years of joint collaboration
• Thousands of hours of testing

Dell EMC VxRAIL

For those customers who are already VMware customers, this would be the most logical fit. The Dell EMC VxRail Appliance provides mission-critical data services at no additional charge with this appliance.

Dell EMC Recover Point for VMs and VMware vSphere Data Protection are built directly into the appliance out of the box. There are also options to add the Dell EMC Data Protection Suite for VMware and Data Domain Virtual Edition (DD VE).

This set of VxRail Appliances come in multiple versions, you simply pick the needs and they are as follows:

• G Series General-Purpose (hybrid use cases)
• E Series Entry-level (small / remote deployments)
• S Series Storage dense appliance (capacity optimized for workloads like Big Data,
  Analytics, SharePoint, Exchange)
• V Series VDI – optimized graphics ready appliances (supports up to 3 graphics
  Accelerators)
• P Series Performance intensive applications (these are designed for heavy
  workloads)

The XC Series, as with all of these turnkey appliance sets from Dell EMC, allow you to increase performance and/or capacity one node at a time, even across multiple generations.

The big win with this framework is the automation and system-wide monitoring provided from Dell EMC. There is a version that is “VMware specific” in the VxRail and VxRack SDDC versions.
Depending on your needs, by adding an OPEN version of designated Hypervisor to this hyper converged appliance set, Dell EMC takes away the guess work and allows you to consume on pay-as-you-grow flex rate. This system allows multiple hypervisors and can even load directly to the bare hardware.

Dell EMC VxRack FLEX

If you are looking for flexible, there is not really another Solution like VxRack FLEX. This deployment type provides VxFlex Manager to control the entire stack. The turnkey deployments are standardized and repeatable and easily extensible. This VxRack Manager greatly simplifies operations.

Heterogeneous

Hypervisors, OS, bare metal containers – but even better, you can use them all at the same time. Choice of factory-installed hypervisor; Microsoft® Windows Server 2016 with Hyper-V; Microsoft® Windows Server® 2012 R2 with Hyper-V; Nutanix AHV; VMware® ESXi™ 6.5 and 6.0.

Scalable

This is one of the most unique pieces of this implementation. While most “stacks” (including the VxRack vSDDC, VMware, Red Hat etc) scale either Up or Out. If look at a virtualized stack today, you’ll probably see only one hypervisor per hardware stack, and generally only configured to run specific use cases. What if you didn’t need to do that? Enter, VxRack FLEX. Add nodes based on CPU Focus or Storage Focus and allow the hypervisors to utilize the pool of resources as they are needed. Increase performance and/or capacity one node at a time, even across multiple generations. Includes VxRack FLEX manager for automation and system-wide monitoring.

Dell EMC VxRack Software Defined Data Center (SDDC)

This solution comes as a turnkey solution with VMware Cloud Foundation which means it includes VMware vSphere, NSX, vSAN and SDDC Manager as its base.

Now, with the ability to extend directly into the cloud and manage VMware in the cloud on multiple vendor’s platforms, it extends the turnkey attraction of this solution.

Pre-configured, pre-loaded, pre-tested, and fully optimized IT stack, delivered as a fully assembled and supported system VxRack SDDC has a base configuration that is a minimum set of compute and storage components, as well as fixed network resources. These components are integrated within one or more 28-inch 42 RU racks.

Within the base configuration, the following hardware aspects can be customized:

Hardware

• Compute
• Network
• Storage
• Management

Minimum set

• Eight 1 RU 1 Node enclosures
• Two Switches
• No minimum. Storage is based solely on the associated enclosure
• VMware Cloud Foundation Manager

VxRack SDDC is a scale-out solution that allows you to add up to 96 nodes with various CPU, memory, and drive options.

Finding the Best Way Forward with Storage Needs

In the case of storage deployments, options designed to simplify IT operations can still present underlying complexities that require careful planning. Effective storage planning benefits from external knowledge and expertise, especially in regards to estimating the proper sizing and prerequisites to ensure optimal outcomes in the shortest amount of time.

IMPRES and Dell EMC use a series of storage and solution calculators for sizing and TCO in order for organizations to better determine whether or not to use Ready Nodes, Appliances, vRacks vs. build their own converged infrastructures.

But even these may require more background knowledge about how the architecture works and how it uses underlying hardware. This includes knowledge on how organizations should best use these solution storage capacities saving features like deduplication, compression or erasure coding.

At IMPRES, we are experts at working with federal customers to assess the current and future storage and performance needs for their data center environments. We work closely to help them determine the best with storage solution.

IMPRES conducts early needs assessments with our federal customers. These assessments may begin, for example, with a comprehensive set of questions surrounding:

• Status of current physical and virtual servers (how many, what kind)
• Storage (current amount and disaster recovery RTO/RPO details)
• Network (speeds, bandwidth, VPN, firewalls, switches, routers, etc.)
• Applications (BI, ERP, CRM, workloads, mobile device management, etc.)
• Operating systems (types, versions)
• Current cloud focus (types in use, use cases)

Information from these assessments offers a great springboard for further discussion and exploration regarding the merits of one type of vSAN deployment vs. another.

However you choose to move forward, we are happy to help your organization work through the specific growing storage sizing and implementation details needed to ensure a successful outcome.

If you need help determining the best way forward with converged infrastructures and storage technology, why not contact us and see how we can help?

We have answers to your questions and access to a wide range of solutions to meet your needs.

About IMPRES Technology Solutions

IMPRES is a trusted partner to all federal agencies. As a leading IT solutions
provider, IMPRES Technology Solutions, Inc., combines decades of technological expertise with a thorough understanding of the federal procurement and contracting process to implement the best possible IT solution for your agency while reducing risk and lowering costs.

IMPRES provides integration and engineering services that emphasize the security, longevity, and capabilities of each deployment. As an SBA-certified HUBZone, we have a lengthy history of high quality contract performance, and have been recognized as an ISO 9001:2015 certified firm. We take pride in presenting innovative and customized solutions to fulfill the technology needs of our customers while exceeding expectations.