Application delivery controllers (ADCs) have been integral in managing traffic and helping organizations avoid outages. In fact, they’ve been around for decades as legacy hardware solutions, but now new software-based ADCs are being developed that are making the ADC even more adept at preventing outages—even on days like Black Friday and Cyber Monday, where spikes in traffic can be near impossible to manage.
Unlike legacy solutions, they work in any environment and don’t require companies to purchase dozens of extra servers to sit dormant most of the year waiting for the need to be used on these infrequent occasions, a proposition that is incredibly cost-prohibitive.
ADCs provide a huge amount of value to websites, e-commerce stores and API back ends and were traditionally under the purview of the IT department. But DevOps teams are now utilizing the ADC to take a more holistic approach to protecting applications, ensuring the efficiency and security of their mission-critical apps.
In this timely eWEEK Data Point article, Dave Blakey, co-founder and CEO of Snapt, offers some industry information about what DevOps should consider with the bursts of traffic we can expect to see on Black Friday and Cyber Monday. Snapt provides load balancing, acceleration, security and caching for websites, applications and services.
Data Point No. 1: Know your server load limits
Having a keen sense of the utilization on your servers is critical. It is incredibly hard to recover from downtime. Users start refreshing, queues spiral out of control, and downtime gets much worse. Ensure you are monitoring the simple details on your servers like CPU, memory and more.
Data Point No. 2: Profile HTTP response times
Use an ADC to profile your servers’ response times. Set alerts for when response times go above what you expect, allowing you to proactively predict overload situations. If your site typically replies in 250ms, set a limit at 300ms to get an alert. It’s common to find specific servers that are misbehaving as well, and if you react fast enough you can avoid issues.
Data Point No. 3: Monitor database performance
Make certain that you have detailed monitoring on your websites dependencies, especially database servers. They are quite easy to profile—monitor queries per second, slow requests and average request times among other metrics.
Data Point No. 4: Use efficient server offloading
Ensure that you are offloading compute-heavy tasks from your servers as best as you can. For example, move SSL termination and page acceleration tasks to your ADC and set them to cache static items. A fully configured ADC should be able to reduce object requests by 50 percent to 80 percent on your back-end servers.
Data Point No. 5: Monitor HTTP status codes
Having a central ADC also allows you to monitor the replies going to clients. Ensure that you are monitoring for increased errors rates at Layer 7, not just Layer 4. Set alerts for 5xx error codes going back to users to stay ahead of any potential issues.
Data Point No. 6: Make sure billing is scalable
Ensure that your credit card processing is scalable. Consider an ADC for load balancing and scaling critical resources, and check to make certain you have benchmarked them.