Cloud-based SaaS solution providers (like Druva) have embraced the public cloud to perform complex and compute-intensive tasks efficiently, securely and at scale. Today there’s more choice when it comes to public cloud architectures, offering greater flexibility to solution providers and their customers.
Up until the recent past, cloud-ready enterprise solution providers have had limited choice when it came to public cloud platforms, and the overall landscape has been fuzzy at best. We had AWS cloud, Google cloud, Azure, Rackspace, all of them seeming to offer different services. Most of them had compute in common, but even there, Google’s App Engine approach was different than just raw compute. Today, the ecosystem has evolved into a set of de-facto services provided by a multitude of IaaS vendors.
In The Cloud, Everything Is Distributed
As a technology company whose customers are reliant on our ability to move and store large volumes of data on a daily basis, Druva saw the importance of leveraging the highly elastic architecture of the AWS stack. We also recognized the importance of being agnostic to the underlying infrastructure, with portability and flexibility being the key to addressing the robust needs of enterprise customers. We built the inSync data protection service on top of cloud services like object storage and distributed database instead of the POSIX services of the on-premise world. Druva’s time indexed, deduplicated file-system stores the file data in object storage and the metadata goes to a distributed database. This means we can run the data protection service on any cloud infrastructure so long as infrastructure provides these services and meets the scalability requirements of inSync customers who are large companies with hundreds of thousands of employees.
Microsoft Azure has matured into a very viable infrastructure that meets the scalability, elasticity, security, durability, and availability requirements necessary to meet our global customer’s enterprise SLAs. For Druva to embrace a new public cloud infrastructure like Azure required assessing the existing components and capabilities, and ensuring a seamless experience regardless of customers storing data in an AWS or Microsoft Azure region.
How We’re Porting to Azure
When inSync runs on Azure platform, the file data is stored in Azure Blob Storage. For the metadata, we use a homegrown distributed database. Why our own database? We needed a disk optimized nosql database that would scale to hundreds of thousands of transactions per second. Other vendors like Twitter also built their own distributed database to address latency and performance needs. Most existing noSQL dbs are mostly memory optimized, and hence don’t have the right cost-performance ratio. Druva’s distributed database runs on top of Azure Premium Storage, the SSD block storage offering from Microsoft. Running on top of SSDs allows our database to scale to thousands of IOPS at millisecond latency.
The Growing Marketplace for Public Cloud
What does is mean to our customers? Druva inSync offers the same set of services regardless of the platform it runs on. But, Druva customers can now choose where their data should be stored and choice is always good. Here’s why:
Stay tuned for a deeper dive into our distributed database architecture in a future blog post in our Tech series.
Curious to see our data governance dashboard for IT administrators in action? Take a trial today.