Photo by Quino Al on Unsplash

Analyzing tricky database problems

Set Theory

Article from ADMIN 44/2018

By Greg Vernon

Database administrators at the Swiss foundation Switch were increasingly encountering corrupt data, which prompted a painstaking bug hunt. The solution they found increased performance by a factor of six.

Troubleshooting complex cloud infrastructures can quickly look like finding the proverbial needle in a haystack. Nevertheless, the effort can be worthwhile, as shown by a practical example at the Swiss foundation Switch. The foundation has been organizing the networking of all Swiss university IT resources for more than 30 years [1]. In addition to Swiss universities, the Switch community also includes other institutions from the education and research sectors, as well as private organizations (e.g., financial institutions or industry-related research institutions).

Because of its high degree of networking, Switch depends on a powerful cloud infrastructure. The foundation provides tens of thousands of users with the Switch Drive file-sharing service, which is based on ownCloud, is hosted 100 percent on Swiss servers, and is used mainly for synchronizing and sharing documents (Figure 1).

Figure 1: Overview of the Switch server infrastructure.

Switch Drive is essentially an ownCloud offering under the Switch brand that manages 30,000 users and about 105 million files. The oc_filecache table contains about 100 million lines. To manage this high number, six MariaDB servers [2] are used in Galera Cluster

...

Use Express-Checkout link below to read the full article (PDF).