Disclaimer: This is a sponsored post and you’ll probably see the content elsewhere on the Internet. Tintri provided no editorial input and the words and opinions in this post are my own.
I recently had the opportunity to speak to Brock Mowry (CTO of WHOA.com) about the company’s experience adopting Tintri in their environment. You can read the case study on Tintri’s website, but sometimes it’s nice to get a perspective straight from the source. If you don’t know of WHOA.com, they were established in 2013 and deliver a “[c]ybersecure cloud hosting platform with an emphasis on compliance workloads, [including] HIPAA regulation and PCI regulation”. They have a data centre presence in Las Vegas, NV and Miami, FL and plans to expand that footprint.
I asked Mowry what one of the main challenges was as a growing cloud service provider and he said “[s]torage was one of the challenges”. The problem, it seems, was when they looked at how much time they spent on keeping the environment running, there was a lot of operational overhead with their storage platform, and they “didn’t want to be scaling by head count – [they] wanted to scale by technology”.
What solutions did they look at?
According to Mowry, at WHOA.com they “optimise [the] network for NFS traffic and get really, really good results operating NFS in [the] infrastructure. Again, Tintri being an NFS-based platform, it was really an easy choice from there”. The benefit of deploying an IP-based storage solution was that they were “able to eliminate an entire fibre channel fabric within [the] data centre”. The added benefit of this was that they were able to reduce the number of “employees that are required to operate that platform. That’s a huge cost saving for [them] because at the end of the day head count is typically one of the most expensive things to operate a cloud infrastructure”.
Why not look at hyperconverged solutions then?
It turns out they looked at a number of hyperconverged vendors, including solutions from Nutanix and Cisco. At the time they ran across a problem with the converged nature of the resources in hyperconverged environments. Mowry provided an example where there was a “need to increase […] CPU and RAM capacity to meet a customer’s workload. Well now I’m sitting on a bunch of excess storage that I really don’t want to power, I really don’t want to cool, and I really don’t want to manage, because it’s not needed”. Note that a number of vendors now offer solutions to that problem, with “storage-only” nodes being available to counter the requirement to scale memory, CPU and storage in equal amounts. At the time, however, Mowry felt that it was best to go with what he describes as a “broken-out” architecture, where they “have storage arrays or storage appliances and [they] have UCS blade systems so [they] can increase RAM and increase CPU to the customer’s workloads without having to scale out our storage at the same time where it might not be used”.
Why go All-Flash?
WHOA.com have deployed both All-Flash and hybrid arrays, because, as Mowry points out, they “have customers who are demanding that lower tier. And a lot of times they’re trying to hit a price point, they’re not trying to hit a performance point”.
WHOA.com are obviously very happy Tintri customers, but not simply because the Tintri arrays they’ve deployed give them per-VM control or nice APIs to use with their own products. Vendors often focus on the technical advantages of the solutions they sell, because they think that’s what demonstrates value to their (potential) customers. But discussions around decreasing operational overhead and improving configuration simplicity by removing fibre channel fabrics are real world examples of how businesses can, in some instances, save money and improve their bottom line by choosing an architecture that aligns well with their operational strengths and experience. People are normally the most expensive part of any type of managed service, so if you can deploy efficient systems that don’t need a lot of people to run them, you’ll be in a good place.
Of interest also was the decision to continue with a decoupled infrastructure architecture that provided them with a solution that scales the way they want it to. In my opinion this a great example of a business choosing a solution that suits them because of a number of reasons, not all of which are technical. Customers like WHOA.com provide a great example of how to understand your requirements (from both a technical and financial perspective), understand your market, and work to your strengths. You can download a full transcript of my chat with Mowry from here.