For me, one of the big attractions of Azure is the ability to scale an application architecture over many data centres around the world. Cosmos DB, Microsoft’s globally distributed, multi-model database service stands out as one of the easiest databases to scale globally.
‘Cosmos DB allows you to add or remove any of the Azure regions to your Cosmos account at any time, with a click of a button. Cosmos DB will seamlessly replicate your data to all the regions associated with your Cosmos account while your application continues to be highly available, thanks to the multi-homing capabilities of the service. For more information, see the global distribution article.’
Many Cosmos DB customers would have chosen to use the product for its scaling benefits and many would have chosen to geo-replicate their data across many Azure regions (data centres). But it might not be apparent that auto-failure is an option, not a default.
For geo-replicated, single leader Cosmos DB accounts it’s recommended that the auto-failover option is enabled. Further details can be found here.
When an enterprise has a large number of engineering teams with many Cosmos DB accounts, it can be difficult to audit the accounts and ensure best practices are being maintained. Azure Policies provide a good way to keep on top of things and ensure a consistent approach to site reliability engineering. My co-worker and I have contributed two new policies to Microsoft’s policy samples repo. The ‘audit-cosmosdb-autofailover-georeplication’ policy will generate an audit report, highlighting single leader Cosmos DB accounts that haven’t enabled auto-failover.
IP whitelisting reduces the attack surface of an asset by limiting the range of IP addresses that can establish a network connection with the asset. Cosmos DB accounts can be configured to enable IP whitelisting, protecting the account and the data within from unauthorised parties. IP range filtering is configured by Cosmos DB’s ‘firewall and virtual network’ controls.
Again, within a large organisation it’s difficult to ensure that all production accounts have the right level of security precautions in place. So, we created the ‘ audit-cosmosdb-ip-range-filter’ Azure Policy to produce a report detailing any accounts that haven’t yet specified an IP range filter.
Large Azure estates with many subscriptions and numerous resources are difficult to manage. Policies provide the tools which allow resource governance to be applied across an evolving cloud infrastructure landscape.
Gary Strange is a Lead Data Engineer at ASOS. He works with 11 teams advising and enabling architects and engineers to design and build ASOS microservices and data analytics architecture.
Sotiris Karras is a Senior Data Engineer at ASOS, consulting architects and engineers in designing and building data-oriented microservice solutions.