Defining performance is key to cloud reaching its potential

A recent report from research firm Gartner has identified that very few IaaS providers are able to offer SLAs for compute or storage performance. The report, Gartner’s Magic Quadrant for Cloud Infrastructure as a Service, aims to establish a graphical competitive positioning for leading players within the IaaS market.

Groucutt states that in order for cloud services to move on from testing and development to widespread adoption for back-office and tier one applications, it is imperative that market leaders start investing in purpose-built technology that guarantees performance, providing businesses with the necessary reassurances they need:

“The Gartner report highlights the biggest players within the market. Whilst it is alarming to see that very few of those have SLAs in place for compute or storage performance, it is unfortunately not surprising.

“The simple truth is that most IaaS providers don’t have the ability to guarantee the necessary SLAs for storage. These service providers operate multi-tenant platforms, where the ‘noisy neighbour’ effect is a real issue. Shared resource necessarily results in shared performance. Service providers need to be able to manage that shared environment to provide consistent, reliable performance. If they don’t, if one user does something on the shared platform that uses a lot of resource, it has a knock-on effect on the service for everyone else on that platform.”

Groucutt continued: “Another issue is the difficulty in separating storage and performance from one another. The performance you get when using spinning disk drives is determined by the size and speed of the disk – in order to increase performance you need to add more disks. This leads to further complications in guaranteeing SLAs.

“As more and more businesses look to migrate their tier one applications such as ERP, accounting, billing and email to a Cloud Service Provider (CSP), guaranteed storage performance becomes much more important.

“Innovation is needed to drive change. Previously, computing environments were used by just one business at a time. Now, multiple businesses are running on the same shared platform which has its own unique set of challenges. This is why we’re seeing the reengineering of so many technologies as well as a surge in purpose-built technologies for multi-tenant cloud providers. Service providers using those technologies can consequently support high performance with guaranteed quality-of-service to all customers on a shared platform.

“With traditional storage, performance is tied to capacity. In a cloud environment, this causes difficulties in guaranteeing performance-based SLAs and it is often necessary to dedicate specific disks to an application. This is a problem for service providers. In order to give a customer the IOPS they need, they need to over-provision storage. This either comes at a cost to the customer or to the service provider, meaning that either the cost of the service is not competitive, or it increases the service provider’s costs. Purpose-built technologies for multi-tenant cloud providers allow storage performance to be separated from storage capacity, allowing for exact levels of IOPS and performance to be defined and allocated to individual users of a shared platform.

“If cloud computing is to take the next step and give businesses the reassurances they need in terms of storage and compute, these innovative technologies can play a significant role in allowing CSP’s to define the necessary performance SLAs,” Groucutt concluded.