Is WAN optimization a must for doing business over the cloud? What else should be in your WAN-cloud performance tool belt, if it isn't already?
Mark Day: WAN optimization is pretty much a necessity, in the sense that the kinds of performance issues that arise because of bandwidth constraints and latency issues are very much the same, whether you're talking about the cloud or whether you're talking about an enterprise WAN.
The problems sometimes get a little worse in that the cloud provider may not be transparent about where the servers are that you're using and whether the servers are changing location over time…. You may be surprised to discover the latency to a given server has increased over time, even though nothing else about the server has changed. With that in mind, WAN optimization is a technology to bring to bear in order to fix the issues that crop up.
For many cloud services, aren't you at the mercy of Internet performance? How can WAN managers ensure that resources are always available when they need them?
Day: It's not that different from … an enterprise WAN setting. You certainly have differences in terms of there being a shared infrastructure rather than a dedicated infrastructure, so there can be concerns associated with that. Depending on the application, that can be the kind of thing that causes people not to migrate [to the cloud].
On the other hand, it has to be said that the general sense of people [who have completed] that migration is that it has been good. It's one of those things where you can certainly see where you can construct an example in which the public Internet is not going to be a good choice. The challenge is not to take those examples and say, "The public Internet is never a good choice."
Generally, we hear a lot about the problems that 'chatty' apps cause over the WAN. Does entering the cloud into this equation complicate or change anything here?
Day: It's basically one of those things where the answer to that question gets complicated because there's often no clear separation between people moving to a cloud deployment and people moving to a different version of an application. So what you'll sometimes find, particularly when people are moving to a software-as-a-service model, is that, in the migration to the cloud, they have effectively gone for a kind of application that's [inherently designed to be] less chatty over the wide-area network.
That's definitely a case you could point to and say, "The cloud fixes chatty application problems," but what [the statement] is missing is that there was actually a rewrite…. You definitely find that a cloud in itself does nothing to fix chattiness, and the increased WAN distances [can worsen latency].
People are starting to step away from the steamroller theory that the cloud is just going to crush everything in front of it.
Mark Day Chief ScientistRiverbed Technologies
There is much in the media about why people are excited about the cloud—it saves money and resources, in terms of people and hardware. But have you heard complaints?
Day: There's definitely a developing backlash. It just has to do with the general problem that something new and being promoted tends to get overhyped by people who don't understand that there are limitations and drawbacks. There are some glitches, some hang-ups, some hangovers and some horror stories.
People are starting to step away from the steamroller theory that the cloud is just going to crush everything in front of it. At Riverbed, we're very comfortable with that middle-ground position [of a gradual migration]. We give people the tools so they can work across the wide-area network and the cloud and be comfortable.
If you're thinking about subscribing to a software-, storage-, infrastructure- or platform-as-a-service model to execute over the WAN, what should you look for? What questions should you ask the provider?
Day: The most important piece is understanding what your organization's business requirements are and having some notion of what those would translate into in the cloud world. It's important to have some notion going into it of what kind of problem you're going to solve. What's going to count as success?
That said, I think some of the things that people maybe don't know they should look for but probably do matter have to do with issues relating to the availability of the service and the SLA [service-level agreement] associated with it. Also, I think it's important for people to actually do trials to understand what kind of performance they're getting out of a certain cloud provider.
Continue reading part 2 of this Q&A: Got experience running a WAN? Cloud computing may not be so scary