Why You Need to Understand WAN Optimization With AWS Direct Connect

With Amazon’s Direct Connect announcement last week, it seems we are going back to the past, when dedicated WAN lines connected our computing resources. Way back, like to when IBM’s SNA was around in the late 1970s and 1980s and when corporations purchased communication lines at the whopping speed of 19 kbps. Now Amazon can connect one and 10 Gbps networks to their cloud.

There is a catch, of course. Your racks of servers have to be connected via one to three initial providers supporting Direct Connect, and if you don’t have this equipment in one of their points of presence, you will have to buy connectivity to get your packets there. Remember, we are talking communicating outside of the Internet on a private network.

One of these providers is Equinix, who operate 90 high-performance data centers around the world. In their announcement, they provide a selection of use cases including:

  • High performance integration with core-internal applications – This allows customers to treat AWS instances as part of the data center LAN and integrate cloud strategy into more applications.
  • High volume data processing – This allows customers to efficiently move much larger data volumes into and out of AWS.
  • Direct storage connectivity – This allows for more regular migration/replication of data from AWS into customer-managed storage solutions.
  • Custom hardware integration – This allows for better integration of custom hardware solutions into the AWS workflow, with real-time streaming of data from/to AWS.

“Cloud access doesn’t necessarily equal the Internet, and in many cases the Internet isn’t appropriate,” says Joe Skorupa, a Gartner research VP on data center issues who has been arguing the need for dedicated links for several years. “First, performance is variable and there are times when a business critical app can be at risk. Second, you don’t have control over QoS if you are running multiple services. And service level agreements around a dedicated point-to-point link will be more expensive but there are guarantees around restore times which should be faster. If latency is a problem, then WAN optimization may be very appropriate.”

So if you are interested in Direct Connect, your first step is going to be to examine what kind of WAN optimization technology to use to move your data across these private networks. This is because your are charged for every packet, so if you can cache and avoid sending some data over your WAN link, you can save some cash, and up your performance too.

There are many vendors in this space, including:

David Greenfield’s Network Computing column earlier this spring

talks about some of the issues with actually getting the kind of performance improvements promised by these vendors. He says, “With the ability to save real dollars by improving WAN connections, WAN optimization vendors are encouraged to tout ever greater levels of performance. But those numbers may have little bearing on your reality.”

WAN optimizers work in a variety of methods to speed up data transfers: by acting as proxy caches to deliver frequent-asked content, by putting in place application-specific techniques to handle Web and database transactions, and by looking at improving overall quality of service performance.

Typically, WAN optimizers have to occur in matched pairs, one at either end of your data link, to ensure the best performance. Blue Coat has figured out a way to only require a single appliance at your end of the link, which comes in handy if you are trying to up your throughput to the cloud.

You can read more of our coverage of Amazon’s announcements from last week here.

Note: I have done some occasional consulting work for Blue Coat, including most recently a video on their MACH5 product.

Facebook Comments