RMS develops large, complex catastrophe models to help the insurance and financial industries understand risk and prepare for the financial impact of catastrophes. Working with highly detailed models and tons of data in regulated industries, our cloud strategy is crucial.
Four years ago we began a transitional journey to move our traditional on-premise based software offering to a cloud-based SaaS solution to one of the biggest cloud service providers. It proved to be a great virtual runway for us. We saw some important quick wins in the public cloud—the major decrease in development time being a good example. But it didn’t take long before the limitations of the public cloud came to the forefront.
There were the basic economics. Gartner and other analysts agree that there’s a point where the cost of subscribing to a cloud service exceeds the cost of an internal private cloud. That definitely was the case for us. The subscription spend quickly surpassed the combined CAPEX and OPEX of in-house operations.
Second was the public cloud’s inability to deliver flawless, predictable, peak performance. Trying to handle our write-intensive, heavy compute load without lag, latency or downtime proved to be impossible using shared resources in the public cloud.
The third focused on shared resources. Insurance is a heavily regulated industry, and our customers’ auditors demanded a more tightly governed environment than could be provided with a shared service.
Faced with this we began our second transition—with help from Kovarus, a cloud consulting firm —from a public cloud to a private one. Our in-house platform comprised of Cisco UCS as the compute platform, VMware vSphere as the virtualization platform, and four EMC VMAX 40Ks handling all our storage. The VMAX is a powerhouse. During the selection process we tested a range of storage options under full-load conditions, including all-flash arrays. Our write intensive load proved that an all-flash solution was not an option from both an economic and performance perspective. A powerful five nines hybrid array with a substantial cache and automated tiering such as EMC’s VMAX, proved to be the best solution for our use case.
Our performance has improved by an order of magnitude and the cost of the environment has been cut 50 percent. That’s just dramatic, isn’t it? Ten times the performance at half the cost by bringing our cloud in-house. And it’s reliable performance. Testing at full load and we haven’t experienced any of the interruptions we had with a public cloud. A lot of the credit for that goes to the VMAX—a true “turn it on and forget about it” storage system. Security concerns for our customers have been answered. Now, in keeping with their own internal standards, our customer’s auditors can get answers to critical questions like “which rack and which servers house our data”, “is there encryption”—which was impossible to audit with any certainty in a service cloud.
I was given the kind of challenge any technologist wants—lead a team in a transition to a new environment, improve performance, cut costs, all in a way that delivers business value to our customers. EMC played a played a major—in many ways an indispensable—role in making that happen.
RMS In CIO Magazine: How RMS Put its Catastrophe Modeling Software in the Cloud
source : emc.com