Cloud Drives Need for Improved WAN Optimization

Arthur Cole
Slide Show

Deploying Applications in the Cloud

While there's more talk than actual use of cloud computing in the enterprise, a Zeus Technology survey looks at the beginnings of a major shift under way. Clear expectations and planning can improve your experience and near-term success.

If the public cloud is in your future, there will be no getting around the fact that application performance will very much depend on your ability to negotiate public networks.

That problem is that once your data has left the confines of your internal infrastructure, you are at the mercy of whatever traffic conditions happen to be in place at any given time. For many organizations, that leaves too much to chance, which is why interest in cloud-ready optimization technologies is at fever pitch.

Riverbed recently upped the ante in this field with a new version of the Steelhead appliance designed to integrate directly into popular cloud services like Amazon's EC2. The goal behind the Cloud Steelhead is to provide a single platform on which applications can maintain peak performance regardless of whether users are utilizing local or cloud resources. The system works in conjunction with the new Whitewater appliance designed to perform the same function for storage.

According to Riverbed executives, the overriding goal is to maintain the flexibility and scalability of cloud environments while bringing down the costs. Added touches like a deduplication engine in the Whitewater appliance are designed to keep data loads to a minimum even as they enhance the platform's ability to support popular cloud applications like backup and recovery.

In fact, B&R is likely to be a primary mover of cloud-based WAN optimization, as the experience of communications firm Rhea & Kaiser shows. The company was having significant trouble replicating virtual machines in the cloud before it deployed NetEx's HyperIP software. The system bumped its long-distance transfer rates from a latency-inducing 10 Mbps to upwards of 800 Mbps, providing smooth connectivity for the firms virtual private network (VPN) and virtual tape network (VTN).

However, some studies indicate that WAN optimization may be of more benefit to some organizations than to others. Orange Business Services, a unit of France Telecom, recently surveyed top executives at more than 500 multinational corporations across Europe and found only about half had any form of optimization technology in place, even though the need for strong application performance in cloud environments was rated as a top priority. The discrepancy tended to fall along regional lines, with many organizations reporting less of a need for optimization provided they had low bandwidth costs and advanced access technologies at their disposal.

If the use of cloud services increases as quickly as the experts predict, it probably won't be long before optimization becomes an integral part of the picture. Public networks offer a vast amount of bandwidth, but as cloud deployments increase, more and more application data will flood the system. As a technology that can both lower costs and increase performance, WAN optimization will seem too good to pass up.

Add Comment      Leave a comment on this blog post
Nov 12, 2010 2:02 AM Marc Le Maitre Marc Le Maitre  says:

I fully agree with you about the future role of wide area optimization in cloud. I'd like to add a new dimension to your thinking.

For 15 years or so we have been virtualizing the LAN and now VLANs play a central roll in cloud architecture. So why not then virtualize the wide area as well and further extend the benefits of virtualization to the wide area networks that connect the distributed components of the enterprise cloud?

Enter; virtual wide area networks (VWANs)

VWANs make both ends of a wide area network appear to be on the same Layer 2. They are built using software-based virtual NICs. VWAN client software is installed on VM instances in the cloud and VWAN server software is installed on the enterprise LAN. The resultant architecture looks like a distributed virtual switch straddling the wide area.

When a VM instance containing VWAN client software spins up in a remote cloud its first action is to create a secure tunnel from the VM through the host machine back to the VWAN server. The server gives the client an IP address from the enterprise address space which the client then allocates to the VM IP stack. The VWAN client may also be given a VLAN association. The VM now behaves in every respect as if it were on the enterprise LAN. Because the enterprise administrator controls the VM's IP address and VLAN association independent of the public cloud provider the VM can be moved from one cloud vendor to another or from private to public cloud sites with zero re-configuration. Regardless of its location access to the VWAN-enabled VM is via the same domain login that it would have used if it were physically on the enterprise LAN.

Once connected the VWAN client and server work together to optimize the WAN between them. They constantly measure the WAN's performance metrics, shape traffic through deep packet inspection and admin policy, bond multiple WANs together for bandwidth and resilience when needed, secure, measure and report the WAN performance allowing the enterprise administrator to verify if SLA's are being met.

VWANs are transparent to Layer 4-7 optimization which means they can be deployed without impacting legacy systems. VWANs are designed to be totally under the control of the LAN administrator and require no provisioning by 3rd party providers. They can run over existing managed connection such as MPLS but do not require them.

Virtualization of the wide area network connects the distributed components of hybrid cloud back together on the same Link Layer under the control of the LAN administrator and answers many of the control, security and performance questions about distributed cloud architecture.

Your thoughts?

Nov 12, 2010 11:54 AM Mark Weiner Mark Weiner  says:

Interesting announcements from Riverbed.  I certainly agree that now is the time for traditional and advanced IT services to begin moving into the cloud, vs. remain as 100% premise-based.  However, there's more than one way to skin a cat, even with something as cutting-edge as cloud storage or acceleration.  One question I have about the news is what about using the cloud to deliver both traditional and advanced acceleration between offices, not just between enterprises and the public cloud?  Also, it's great to see more vendors partner to deliver services the way companies need them (via the cloud), but limited flexibility is still an issue, even with the services just announced.  Here's my take on two different approaches to delivering cloud IT services:

--  Mark Weiner, Virtela

Nov 13, 2010 11:52 AM Marc Le Maitre Marc Le Maitre  says: in response to Mark Weiner

Although interesting, one aspect I dont understand of Riverbed's strategy is why they build this function in a hardware appliance. The need for hardware to optimize what is essentially a software service seems to be less intuitive.


Marc Le Maitre -


Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.