Situation
FeedItBack runs a platform that provides customer engagement feedback analytics.
FeedItBack runs a platform that provides customer engagement feedback analytics.
Due to their primary workload, their “accidental DBAs” have been struggling to catch up with thoroughly reviewing the state of the databases their applications use. Some parts of the system were increasingly underperforming, and there was no in-depth performance monitoring in place to expose potential issues. They also developed a new set of functionality that was performing poorly on their test server, and they needed it reviewed and improved before deploying to production.
Shattered Silicon deployed Percona Monitoring and Management to provide insight into performance bottlenecks, and carried out a review and optimization of the most problematic queries and configuration. The queries required by the new functionality were also reviewed and optimized to make their performance much more acceptable for interactive use.
As a result of the work carried out, the interactive performance if the system was noticeably improved, and FeedItBack now have monitoring in place that allows them so spot any potential performance issues before they start impacting the user experience.
Increase your performance, flexibility and business continuity by leveraging our specialists 24/7.
Zolo is Canadian real estate brokerage. One of their most important assets is their website, which lists hundreds of thousands of houses for rent and sale from all over Canada.
They required a thorough review of their database estate in order to ensure that they are getting the best possible performance from their servers and that they are configured in a consistent and manageable way.
Shattered Silicon deployed Percona Monitoring and Management to provide insight into performance bottlenecks, and carried out an exhaustive review and optimization of the queries and the servers’ configuration.
The database server estate was rationalised down to half of the number, and simplified into a primary and HA hot spare arrangement. This made the estate much easier to manage. The overall performance was also improved by a significant multiple, with some queries that were affecting the user experience improved by an order of magnitude. To protect this critical infrastructure, alerting via PMM was also configured to provide immediate warning if an anomaly exceeding conservative thresholds occurs.
Increase your performance, flexibility and business continuity by leveraging our specialists 24/7.
Lumesse runs a recruitment platform called EasyCruit. Their platform allows customers to keep track of candidates and the roles they are being recruited for throughout the recruitment process. EasyCruit also acts as a database of candidates and enables them to search for candidates they engaged with previously and reach out to them for new roles.
EasyCruit was suffering from major performance and stability issues with the underlying MySQL database. Not only had the performance become unusable, but an even more serious problem was manifesting, resulting in tables on the replicated servers to regularly become corrupted.
Shattered Silicon converted all EasyCruit tables to a storage engine more suitable for their workload. The conversion significantly improved performance and concurrency and also resolved the table corruption issue since it was caused by a bug in the original storage engine. Shattered Silicon then proceeded to carry out an in-depth performance review, and delivered changes that improved overall performance further by two orders of magnitude. Finally, the OS and MySQL were upgraded to a newer version that offered a longer amount of time before it’s scheduled support end-of-life date.
As a result of the improvements, EasyCruit performance was improved. Prior to conversion, most pages timed out a majority of the time. After conversion, every request returned almost instantaneously. Lumesse were also able to scale down their server estate from the original 17 servers down to 2, of which the 2nd was there purely as a hot spare in case the primary failed. The primary server of the same spec as the original 17 was able to serve the load using only 5% of its capacity. Overall, we delivered an 8x reduction in costs with a ~100x improvement in performance.
I can’t recommend Shattered Silicon highly enough! They were quick to respond and understand what we needed, knew immediately what action we needed to take and helped us apply their suggestions. They are quick to offer very detailed advice, and have been a real pleasure to work with (and will continue to work with in the future). In terms of value for money the benefit we’ve gained in just a few hours of their time has been fantastic!
Linux Consulting
Linux Support
System Analysis and Configuration
Performance Tuning and Optimisation
High Availability Solutions
Distaster Recovery Solutions
Email: [email protected]
Tel:+44 7740 195136