Visibility is Key to Assuring Application SLAs in the Cloud
More and more businesses – both small and medium-sized businesses (SMBs) and large enterprises alike – are migrating business-critical applications to the cloud in the hopes of achieving lower cost and greater efficiency.
But as new complexities disrupt business and IT, maintaining predictable business operations and a reliable user experience is a significant challenge; the typical enterprise network lacks the visibility, control and bandwidth required to rapidly migrate application workloads into the cloud, prioritize all types of Internet traffic and protect the quality of the user experience for strategic cloud-based applications.
Delivering an “always-on” experience in today’s complex IT environment means understanding the applications that are critical to the business and assuring their SLAs, while being nimble enough to shift gears at any given time. A change in thinking is required to ensure successful cloud migration, according to Brendan Reid, vice president of product marketing at Exinda, a provider of WAN optimization and Network Control solutions.
Exinda, based in Boston, Mass., focuses on midmarket enterprises --which the same segment leading the cloud-migration charge, says Reid.
“Application migration to the cloud introduces another level of variability to the network and is a prime example of the predominant networking trend impacting companies,” Reid explains. “The network is becoming more and more complex – the problems network managers dealt with five years ago were singular in nature and could be addressed by adding bandwidth. That’s all changed because of cloud.”
When it comes to migrating applications to the cloud, the major concerns are loss of visibility into performance and loss of control, according to Jim Rapoza, an analyst with the Aberdeen (News - Alert) Group.
“When your key applications and servers are running in the cloud, they are obviously no longer on your network. This makes it much harder to do end-to-end monitoring and optimization in the traditional ways,” Rapoza explains.
Among the biggest network factors that companies need to consider when migrating to the cloud are service-level agreements (SLAs) and whether the new network scenario will support a company’s application-level SLAs.
“It all comes down to application SLAs – that is all anyone cares about. To enforce or adjust SLAs based on tiers of criticality – where the most strategic applications get the highest priority – is a key factor,” says Reid.
When cloud is introduced as another variable, coupled with the increasing bring your own device (BYOD) scenario, the equation becomes more complex as more employees using applications consume them over the public internet which means the network team now has to deliver the same application SLA amidst very different network conditions.
“The situation is further complicated by recreational internet usage, such as employees using YouTube and Facebook (News - Alert), or streaming sports highlights, while at work.. So while work applications like CRM have gotten more complex, at the same time you also have employees that are using that same network connection for recreational use. But it gets even more challenging in the social enterprise, they could also be using YouTube (News - Alert) for a marketing campaign, so it’s critical that the whole picture is taken into consideration,” Reid explains. “Five years ago, content was king, today context is king.”
Among the biggest issues for companies that have migrated part or all of their applications to the cloud is a slow return on investment, which is driven by a lack of visibility into the network and adequate application user experience testing before moving applications into the cloud.
“Customers tell us that migration to cloud isn’t yet providing the promised returns that they thought it would. The reason is the inability or lack of preparation upfront by companies before they migrate their applications to the cloud,” says Reid. “Companies are moving into the cloud under the premise that they are going to get some kind of cost savings, but in reality, they are underprepared.”
The majority of companies enter into a cloud migration project believing they will immediately reap measurable cost savings. We saw the same thing in phase 1 server virtualization as well as BYOD adoption, according to Reid.
“But like these other initiatives, customers frequently call us part way through a project as they begin to realize they were not really prepared from a variety of perspectives,” explains Reid. “We see it time and time again. For example midmarket customers speak about challenges of migrating applications to the cloud and still delivering their internal SLAs, in particular for., disaster recovery, CRM and unified communications. The resounding feedback is they were not adequately ensuring a reliable and predictable user experience on the public Internet. Preparation is understanding the impact on the network before migrating to the cloud and adjusting SLAs or adding network bandwidth policies before problems arise.”
He notes three typical challenges:
· Challenge 1: The customer didn’t adequately assess the network’s readiness to support critical applications over the public internet. For example, one day their CRM or email was premise-based and operating on a predictable network, and the next day it was fully dependent on the public internet for user experience.
· Challenge 2: A byproduct of the previous point is SLAs – more and more companies try to deliver internal SLAs to the company around critical applications but end up not being able to meet the same level of consistency from the cloud as on premise.
“We hear from companies that the necessary SLA adjustments are often not made – consider the same example where you had an SLA that said your CRM would be up X percent of the time, take X minutes to resolve a problem and provide X milliseconds response time to the user. Now you’re running CRM from the cloud, which means it’s competing with Facebook and YouTube and everything else for bandwidth – if you don’t adjust your SLA or put new tools in place to assure them, you end up not meeting them,” Reid says.
Challenge 3 involves moving applications to the cloud and the way it changes how help desk services are delivered. “You’re now reliant on a third party in the SLA equation in terms of application delivery and problem resolution – it complicates things greatly and needs to be considered more than it is,” Reid adds.
Traditional WAN optimization solutions are, as the name says, designed for the WAN, Rapoza explains.
“They do a great job making sure that traffic and applications on the WAN are running well,” he says. “But once an application or server is in the cloud and no longer on the WAN, those traditional optimization systems are now dealing with a significant portion of critical traffic that they have no visibility or control over.”
Traditional WAN op solutions are multi-sided solutions, Reid says, with many businesses using a two-appliance solution – one in the branch, one in the data center. But that doesn’t always work with cloud.
“You put one appliance in the data center and one appliance in the branch office so they can make traffic in both directions move faster,” Reid explains. “But when you move into the cloud scenario (e.g., Salesforce.com (News - Alert)), that configuration isn’t possible anymore because Salesforce.com is not going to let you put a WAN optimization appliance in their data center. So what do you do? Your choice is to not optimize that traffic and hope for the best, or use a company like Exinda who also offers a ‘one-sided’ or ‘asymmetric’ solution, which means you put one appliance in your branch and nothing in the cloud. That one intelligent appliance understands how to prioritize and optimize web traffic without needing that second appliance.”
WAN optimization is about compressing bits or making bits move faster, but businesses need to take more intelligent approach to migrating applications to the cloud, Reid says.
“The more real-time applications are in place, the more applications are competing for resources. There has to be a differentiation between cloud application control versus WAN optimization, and thinking more in terms of context versus content,” he says.
A WAN optimization solution is inherently a two-sided equation. “In the cloud scenario, you can’t do that. You now have a toolkit problem to deal with. Exinda puts in place single-sided solutions to control incoming traffic from the internet and to provide that control and policy bandwidth allocation,” Reid explains.
Exinda’s focus on SLA assurance is critical to the success of its customers spanning many industries and countries:
? 6 million students worldwide rely on the Exinda Learning Network
? 11,000 branch offices collaborate on the Exinda Branch Network
? 8 of the top 10 hotel brands run on the Exinda Hospitality Network
Over 300 services firms deliver on the Exinda Services Network
? Governments in 31 countries depend on the Exinda Public Service Network
“We provide a level of intelligence to allow companies to assess the impact cloud will have on the network,” says Reid. For example, when looking at CRM on premise versus in the cloud, Exinda helps companies understand and assess the network impact by assessing readiness from a network perspective.
“The company carries this out by helping its customers set up bandwidth-allocation policies to allow bandwidth to be available based on a tier system, to ensure the ones most important to them always get the same bandwidth,” Reid says. “We offer a level of visibility and reporting that allows customers to get very granular.”
Loss of visibility and control can continue to plague businesses that have not adequately prepared their cloud strategy to incorporate not just the volume of network traffic, but also the type of traffic it’s handling.
“One of the problems customers speak about is differentiating types of network traffic,” Reid says. “Without a rich level of visibility, you are not able to distinguish between the internet applications you want to guarantee and those you want to constrain or control.”
The Exinda Network ControlSuite is designed for medium-sized enterprises functioning in today’s rapidly evolving network landscape, which includes private and public clouds; virtualization and consolidation; converged networks for voice, video and data; and a mobile workforce that uses multiple devices to access strategic business applications.
In a classic server and application deployment, businesses can fully monitor and optimize these systems since they are running in the data center and completely on the company network, Rapoza explains.
“Once they have migrated, they now have a major blind spot once the service leaves the company network and goes to the cloud,” he says.
The key metric when it comes to the entire application and network infrastructure lies in the user experience, says Rapoza.
“Everything else is secondary. Even when someone calls IT and says ‘the network is slow,’ they don’t really mean that the network is slow. What they mean is that the application they are using is slow and giving them a bad user experience,” he explains. “In the end, all of an IT department’s performance dashboards can be happily showing green dials that say everything is fine, but if users are having a bad experience, those dashboards don’t mean a thing.”
In a recent survey*, CIOs reported that they were five times more likely to report the user experience was more important than throughput.
“All of these new variables are increasing the complexity of the network,” adds Reid. “In a client-server environment, which is a single variable – you have a bit-centric problem Now we have a business-centric problem. The variables are much more complex and you therefore cannot solve it with a single-variable solution, you have to address a multi-problem user set.”
Cloud is one of these variables making these issues that much more complex. When considering a cloud strategy, organizations need to be able to guarantee a number of things, Reid says:
1) Who is the user?
2) What type of user is it? A guest, an employee, an executive?
3) What application is being used?
4) How strategic is that application?
5) What else is going on in the network at this time?
Taking a Multi-Faceted Approach
While many WAN optimization vendors tend to focus on throughput and reducing the amount of data running on a network, Exinda focuses on one fundamental problem set: application SLAs and structure.
“Whether you are running applications across a WAN or you are consuming apps from the cloud, the one thing that never changes is you have to assure application SLAs,” Reid says. “So focusing on SLAs and not just throughput, you are in a much better position to handle this multi-faceted network environment.”
Exinda helps with SLA applications by managing and maintaining them throughout the migration process.
“We are taking a more realistic approach to what companies’ application environments actually look like today,” Reid adds. “To be able to do that, you have to have a focus on technology and policy, and understand the context by which the applications are being used. You can then set policies so customers can allocate bandwidth to the apps that are important to them.”
Without visibility, an enterprise doesn’t really know what is happening with their applications and performance, Rapoza adds.
“Businesses really need as close to end-to-end visibility as is possible,” he says. “In the end, if you can’t see every part of the application system, odds are that key problems will be happening in the areas where your enterprise doesn’t have visibility.”
Edited by Stefania Viscusi