As a field engineer at Rubrik and a former customer, I have seen data centers (DCs) containing all types of vendors and products. However, I have never been able to build an environment from a blank slate—or what’s known today as “greenfield.” In this post and subsequent webinar, I’ll explore the characteristics of a greenfield data center using two of my favorite companies and their products.

Objectives:

As guiding principles, I’m aiming to do three things in each area of the DC:

  • Simplify: Deliver the most bang with the least overhead
  • Soar: Accelerate business-critical unencumbered
  • Shrink: Represent the smallest physical and environmental footprint

Below, I walk you through all aspects of the DC, incorporating each of these principles to create simple, modern, and scale-out architecture.

Applications and Compute

Technology exists to enable business needs. Without the business, the bits, bytes and architectures are fun, but lack a unifying reason to live. Thus, for our purposes here, I’ll assume we’re cutting and pasting a modern, virtualized environment into this ideal data center. It runs services like Oracle, SQL, and Exchange, and holds terabytes of files and application data that are core to the enterprise.

Following Moore’s Law, the compute layer has become increasingly dense with single sockets rivaling what 4- and 8-socket systems struggled to achieve just 10 years ago.

In this part of my “If I Could…” DC, I would use something like Dell’s FX server series that delivers four 2-socket nodes in a 2U form factor. Each socket is capable of 22 cores and an accompanying 768GB of RAM, which means that a single 2U chassis can bring 176 physical cores and 6TB of memory to bear. Wow!

Since I’m dreaming, but not financially delusional, I’d deploy 2-3 chassis of moderately equipped servers based on the above platform (or similar from HP, IBM, or Cisco), which would consume 4-6U of rack space, 16-24 x 10-gigabit Ethernet (10GbE) ports, and 8-12 amps of power.

Storage

Pure Storage’s FlashArray//m is the logical choice for storage on all fronts with its industry-leading data reduction, unrelenting performance, and user-friendly interface. Just 3U of Pure delivers up to 40TB of native capacity, which transforms into 120TB in most customer data centers (based on average data reduction of 5.5:1).

This efficient use of flash is something easily overlooked when vendors start throwing around numbers and quotes. Whether it’s repetitive instances of test/dev or large shares of corporate documents, strong data reduction means the data layer stays tight while the business grows.

For connectivity here, I’ll go with iSCSI (though I’ve always been an FC guy) to achieve a unified fabric.

Backup & Recovery

Protecting that growth is key to guaranteeing a DC’s success. Rubrik’s Hybrid Cloud Appliance is undeniably the best way to do it. 10-Gigabit Ethernet interfaces and enterprise SSDs open the gates wide to rapid, frequent, and transparent protection of mission-critical operations.

More than backup, though, Rubrik provides the path forward to continuity and test/dev through live mount and instant recovery of virtual machines, locally or replicated, in less than a minute. Can your backup solution bring multiple services like SQL and Exchange online in seconds with performance? Mine couldn’t until I switched to Rubrik.

Another great attribute of Rubrik is the insight it brings to production datasets. Nothing threatens a lean data center like inefficient processes, especially those that churn unique data. Rubrik’s System Capacity report delivers VM-level intelligence into daily growth patterns. This one report can identify trouble spots before they trigger costly infrastructure expansion.

Archive

Rubrik supports many public and private platforms for long-term backup archive, but Pure’s new FlashBlade is one that excites me. Listen to this: 1.6 PB (that’s petabytes) in just 4U of rack space! Say goodbye to racks of power-sucking, hot and heavy near-line disks. FlashBlade can grow as you go and collocate file and unstructured data alongside those yearly backup archives. Nothing nails my “Shrink” objective more than FlashBlade does.

Network

The best way to tie it all together is with two simple 48-port 10GbE switches from your favorite vendor (I know Cisco, but hear great things about all the others). With those in place, I have an incredible data center in a third of a rack.

  • Total Rack Space: 15-17 rack units
  • Total Power: 16-20 amps
  • Total Switch Ports: 32-40 ports (10GbE)

It’s fun to dream, and while few of us can start from scratch today, we all get incremental opportunities to rebuild the future. What’s your wonderland? Join me and Pure Storage CTO Chadd Kenney on Thursday, June 16 for a live webinar during which we’ll discuss this greenfield data center and more. Register now!

Joining from EMEA? No problem! Our colleagues in Europe will be running a similar webinar on July 14th: click here to register for the EMEA event.