As a managed service provider with a hosting operation, how to you prepare for customers with unpredictable web site traffic surges? Here are some best practices from Voxel, the managed hosting and cloud provider.

John Moore

February 17, 2010

2 Min Read
How to Help Customers Survive Web Traffic Spikes

web traffic spikeAs a managed service provider with a hosting operation, how to you prepare for customers with unpredictable web site traffic surges? Here are some best practices from Voxel, the managed hosting and cloud provider.

One of Voxel’s managed hosting customers, Urban Dictionary, was the recent subject of a Facebook craze and a 20x boost in traffic compared with its previous record.  Raj Dutt, Voxel’s chief executive officer, pointed to three offerings that helped Urban Dictionary deal with the influx.

1. Content delivery network

Urban Dictionary taps Voxel for dedicated servers and also uses the company’s CDN, which caches content in geographically distributed hubs.

“The CDN acts as an insurance policy of sorts if they get spikes in demand,” Dutt said.

CDN, he added, in most circumstances, let customers deal with increased volume without affecting the base infrastructure.

2. Cloud computing

A certain percentage of traffic isn’t cacheable, Dutt noted. CDN may be able to offload 90 percent of the traffic. But in a severe crunch, the remaining percentage can significantly affect the main servers generating dynamic content. CDN helped Urban Dictionary with the Facebook onslaught, but needed to provision more servers for dynamic content.

So Urban Dictionary tapped Voxel’s cloud service, adding 20-plus servers to its usual roster of six or seven front-end Web servers. The advantage for the customer, Dutt said, is having the ability to pay for the additional servers for a few days as opposed to buying 20 servers or renting 20 servers for a month.

3. Application profiling

Throwing masses of servers at a demand spike doesn’t always help. Dutt cited the example of Twitter, which bumped into scalability issues in 2008. Dutt said Twitter had an architecture problem that “had very little to do with their ability to add servers.”

The remedy? A hosting provider should take the time to profile a customer’s application to determine its scaling points — the areas in which resources get pinched. This type of service not only helps customers, but helps the service provider stand out.

“Managed hosting companies that have the capability to help customers profile their applications and define those scaling points along the way will really delivery a lot of value to their customers,” Dutt said. “This is a tremendous opportunity to differentiate.”

Application profiling may be within the grasp of service providers — provided they have the professional services resources to dedicate. For most companies, CDN and cloud will be a matter of finding good partners. The key takeaway: your most demanding customers will expect you to leverage a mix of services — your own or someone else’s — to get the job done.

Read more about:

AgentsMSPsVARs/SIs
Free Newsletters for the Channel
Register for Your Free Newsletter Now

You May Also Like