The challenge
SimplePart's database tier was a federation of more than 30 SQL Servers, all read-write, kept in sync in near real time across four datacenters. Behind them sat web farms serving thousands of IIS site bindings, and an SLA of 99.99%. The environment had to be run by a small team, and had to stand up to an annual PCI audit.
What SQLBeat did
One SQLBeat instance, on a single physical server, watched every SQL Server, Windows server, web server and network device in the environment: 106 in all. It read the event logs and key WMI metrics of every Windows server, the performance data of every SQL Server, and the logs of every website, about 44,000 rows a minute, into its warehouse. It also acted as the syslog receiver for network devices.
Its audit module kept a history of every procedure, table, index, login, permission and configuration file across the federation, so drift between servers was caught quickly and change was documented. That record was a key source of evidence in the annual PCI audits.
Because SQLBeat is T-SQL, the team built on it directly: website availability and SSL checks across every datacenter, detection of malicious traffic with live blocking at the firewall, automated regression tests, and SLA reporting for clients, all driven from SQLBeat's data.
The result
A large, complex environment run by a minimal number of DBAs, with critical issues caught and routed in real time, and a complete, queryable history for incident research, tuning and audits.