Reference Architecture, Converged, & Hyper-Converged Infrastructure: A Pizza Analogy

hyper-converged infrastructureThis morning, our CTO Chris Ward delivered an internal training that did a great job breaking down reference architecture, converged infrastructure, and hyper-converged infrastructure. To get his point across, Chris used the analogy of eating a pizza. He also discussed the major players and when it makes sense for organizations to use each. Below is a recap of what Chris covered in the training. You can hear more from Chris in his brand new whitepaper – an 8 Point Checklist for a Successful Data Center Move. You can also follow him on Twitter.

Reference Architecture

According to Chris, reference architecture is like getting a detailed recipe and making your own pizza. You need to go out and buy the ingredients, make the dough, add the toppings, and bake to the perfect temperature. With reference architecture, you essentially get an instruction book. If you’re highly technical, following the recipe is manageable. However, if you are more of a technology generalist, or if you’re newer to the filed, it may be difficult to follow and the chances you get lost in the recipe can be fairly high. The benefit here is that you have flexibility to make the pizza the way you want it. The downside is it doesn’t save you a ton of time. You still need to order the equipment, wait for the order to arrive, and then put it together.

The Players

  • EMC’s VSPEX – EMC storage, Cisco UCS compute, Cisco networking
  • Nimble – Nimble storage, Cisco UCS compute, Cisco (newer offering)
  • FlexPod – NetApp Storage, Cisco UCS compute, Cisco networking

There are several use cases when it makes sense to utilize reference architecture. These include when an organization:

  • Has disparate vendors where converged or hyper-converged infrastructure may not be an alternative and the organization is not open to a vendor switch
  • Requires more flexibility in components than converged infrastructure provides (i.e. you can add some extra garlic to your pizza and not have it be a big deal).
  • Doesn’t have a hardware refresh cycle between storage, compute and networking that is in alignment (i.e. you do not want to double up on servers you just bought last year)

Converged Infrastructure

Converged Infrastructure is like a take home pizza you buy at a grocery store (it’s not delivery it’s Digiorno!). Converged Infrastructure is more prepackaged than reference architecture. The dough has been made, the toppings have been added, but you still have to put it in the oven and bake it. Vendors do the physical rack, stack and cabling at the factory and ship it directly to the customer. Customers can expect this typically in 30-45 days of placing the order. You don’t have to wait months to get all parts shipped and then assemble yourself. However, the infrastructure is set in stone. If you are an IT department with a different shop than what you are getting with the converged infrastructure option, you can’t mix and match. There is also still integration that comes with converged infrastructure.

The players

There are several use cases when it makes sense to utilize converged infrastructure. These include when an organization:

  • Requires fast time to market (typically 30-45 days from order to constructed delivery. Keep in mind there is additional time on the front end before the order when planning the solution out).
  • Is building out of application PODS or private cloud. This is typically more of a use case in the enterprise space. For example, rolling out a new SAP environment and having, say, Vblock be solely dedicated to that one app running on it. Another example is a larger VDI project.
  • Requires known, guaranteed and predictable performance out of the infrastructure. With Vblock, VCE guarantees you the performance that you do not get with reference architecture
  • Requires large scalability – you can add to it over time. Keep in mind you need to have a clear direction of where you are headed before you start.
  • Is stuck in the mud with operations and or maintenance validation tasks. Again this is a more relevant use case in the enterprise space. Say an IT Department needs to upgrade from vSphere 5.1 to 5.5 in a cloud environment. This could take them 3-4 months to do all testing, etc. By the time they get everything together there could be a new update on its way out. This IT Department is always 2-3 upgrades behind because of all the manual work. With converged infrastructure, vendors do that work for you.

Hyper-Converged Infrastructure

A hyper-converged infrastructure is the equivalent to a fine dining pizza experience. You can sit back and have a glass of wine while your meal is served to you on a silver platter. Hyper-converged infrastructure is an in-a-box offering. It’s one physical unit – no cabling or wiring necessary. The only integration is to uplink it into your existing infrastructure. If you choose to go this route, you can place the order, overnight ship it and expect to have it on your floor in 48 hours. This is obviously a very fast time to market. As this is the newest space of the three, it’s a little less mature in terms of scalability. Hyper-converged infrastructure often makes the most sense for midmarket companies. Keep in mind, hyper-converged infrastructure is a take it or leave it, all or nothing deal.

The Players

It makes the most sense to utilize hyper-converged infrastructure when companies:

  • Storage and compute refresh cycles are roughly in sync
  • Are looking for out-of-the-box data protection (Simplivity)
  • Require known/guaranteed/predictable performance
  • Are looking for rack space and power consolidation savings
  • Require a small amount of scalability
  • Want a plug-and-play approach to infrastructure.

Which way makes the most sense for you to eat your pizza?

You can hear more from Chris in his brand new whitepaper – an 8 Point Checklist for a Successful Data Center Move. You can also follow him on Twitter.


Photo credit:

By Ben Stephenson, Emerging Media Specialist