Application acceleration, though maturing at a rapid clip, is still getting its share of buzzword hype. We recently sat down with Ray Nahorniak, director of network solutions services delivery for Forsythe Solutions Group, an IT infrastructure integrator, to discuss who needs application acceleration and what benefits it can bring to the table. Nahorniak also touched on what expectations companies should have when evaluating different application acceleration solutions.
Application acceleration often blurs the line between the application group and the network group. How do the networking folks get involved in this when it's a responsibility that's been held onto pretty tightly by the application folks?
Ray Nahorniak: When you look at application acceleration, it's really a means to provide LAN-like application performance across the WAN. If you think of application protocols that were designed for a LAN environment, when you have those same applications going across the WAN, the network demand is increasing. So if you think in terms of a call centre across the country or internationally being served by a data centre that's located in the U.S., you're going to get delays across the WAN for that application support. Application acceleration is really the means to provide LAN-like performance for that application. I'd say it's more on the network side, but it does work in conjunction with the application. It is the means to improve performance to near LAN-like levels.
Many companies are looking into application acceleration across the WAN and trying to get regional offices and branch offices on board. What are the real first steps a company should take?
Nahorniak: If you want to consider best practices, I would say No. 1 is to define your application needs and priorities. What are the applications that are critical to running that enterprise? Baseline your current performance. Use your application needs to determine the top two or three vendors and ensure that the vendor products will scale according to your projected growth. Perform a pilot trial and measure the results. Then implement the solution and optimize it to gain the best performance.
It's critical to understand there isn't a one-size-fits-all solution. You need to go through the due diligence and follow best practices to make sure you're making the right choice that will give you the most benefit for your enterprise.
What's a realistic expectation an enterprise should have when looking at acceleration tools and appliances? We hear statistics from different vendors quoting certain percentages of improvement. How do companies know where to set their expectations?
Nahorniak: It's important and I think part of best practices to trial solutions in a lab environment using the application. Then, on a smaller scale, use the application you plan to accelerate to see what the level of impact will be within your organization. There are case studies in which I have seen up to 80% improvement, but every organization is going to have a different experience. That's why it's important to trial it. If you think in terms of voice over IP (VoIP), you want to make sure that your quality of service is not going to be affected by your specific choice of application acceleration product. In some cases, it knows which packets to just pass through; they don't go through any type of acceleration and they don't want to be held up in queue for the application acceleration software to look at that and say, "OK pass that through." There is a means to say that this traffic doesn't even get considered and just pass it through.
What are the particular types of applications and protocols that can present challenges when enterprises are trying to accelerate them?
Nahorniak: VoIP applications are not manipulated. If you want to look at the most benefit, think of email. If I send email across my network and I have an attachment to it, and you're sending the email back and forth and people are responding to it, that attachment keeps getting sent across the WAN every time. If you look at that example, application acceleration devices at both ends say: "Hey, I've already got a copy of this. I'm not going to send it across the WAN this time. I know there's an identifier to it." As it goes across, each end reattaches it, so it does get presented back to the user with the attachment there, but it's not because it's going across the WAN again.
Are there any applications or protocols that you know of that really cannot see any benefit with acceleration tools?
Nahorniak: I don't know if there are specifically, because there are different vendors that tailor their application acceleration products to support different applications. For the most part, I think there are benefits. But I have heard some anecdotal information that said there was a trial performed and the results really surprised everyone because it did not provide much of an improvement.
How do companies, once they make a decision to move forward with application acceleration, identify which vendor is right for them?
Nahorniak: Define a list of requirements and say, "Here are my specific applications." It's important to define what those are and then look to the vendor that says, "OK, you're running SAP and Oracle. Yes, we actually have a tailored solution for you that looks at how you can optimize the protocols those applications use." So how you do compression and protocol acceleration is tailored specifically to the applications you're running. It really is looking at the different vendors and the products they provide.
When you're looking at vendors, we're talking Cisco, Riverbed, Juniper, Blue Coat, F5 Networks, Packeteer; and probably Cisco and Riverbed seem to be establishing themselves as the market leaders.
Among all of those vendors, it seems this whole idea of application acceleration is becoming a major buzzword right now. Do you see any changes coming in the market as it continues to mature?
Nahorniak: I know Cisco, as an example, says, "I don't want to hold up any voice over IP traffic with my acceleration box." So they've tried to make sure it's not a bottleneck for that traffic. They look at that in conjunction with other applications to keep it from being held up through application acceleration.
The market is maturing. The market is looking into how to integrate security. I think that's one area: how to make security transparent as far as how application acceleration handles it.
Have you heard any application acceleration horror stories where things didn't go the way they were expected?
Nahorniak: I wouldn't say necessarily horror stories, but I've heard anecdotally where the results were a surprise. I think you can always find a horror story if you find someone didn't implement it right or follow the instructions on how they were supposed to implement it.
I don't want to include the human error, but I would say [there are times] where everyone went in saying this is the right solution for you as an enterprise, let us implement it, let us trial it and we're going to prove to you that it works the way we expect, but it didn't.
What are the misconceptions that enterprise IT departments can find themselves falling victim to when looking into these kind of solutions?
Nahorniak: One misconception is they feel it's too immature. I feel that it's at a maturity level now where companies should be looking at it. If they've looked at it before and have not considered it, there's enough information out there now, there are enough companies using it now that they can get good information to say, "I believe there are other companies I can look at that have implemented it and they're in a similar business environment that I'm in, and it could potentially do a lot of good."
Do you have any other suggestions for companies considering application acceleration?
Nahorniak: These are my three key messages:
Application acceleration should be included when discussing data centre consolidation. Ensure business needs are being proactively considered. So if you're having a discussion regarding consolidating the data centre, make sure you include application acceleration as a topic of discussion. It should be, because otherwise you're potentially introducing impacts onto your network that could be resolved with application acceleration.
Other businesses could be overlooking the benefits of application acceleration. Is your network ready from a performance perspective? If you're not considering consolidation and you have a bunch of branch offices out there, you could be getting a lot of benefit from application acceleration.
The third and most important point is to consider best practices when determining the best application acceleration solution for your enterprise. If you don't plan it right and test it right, I don't think you'll get the results, and that could potentially lead back to the stories where an enterprise comes back and says, "You know what, we implemented it and we didn't really see any benefit with application acceleration." You should do your due diligence right away.