IT Briefcase Exclusive Interview: Database-as-a-Service’s Growing Role in the Enterprise

The rise of Database-as-a-Service (DBaaS) platforms is causing enterprises to evolve as well, leading to new thinking about how internal engineering teams are constructed and how infrastructural needs are met. Peter Nichol, CEO of Instaclustr, offers his point of view on how the adoption of new database strategies is affecting enterprises – as well as the advantages and potential issues to watch out for.

• Q. What factors contribute to organizations opting for a Database-as-a-Service strategy, rather than managing their databases internally?

A. While there are certainly some expected considerations such as operational and infrastructure cost efficiencies, we’re seeing that an increasingly important factor influencing a DBaaS implementation is risk mitigation.

The need for most applications to be continuously available means that companies have to eliminate, as much as possible, the risks threatening continued operations of their infrastructure and database. The complexity associated with operating a truly global scale business and applications means that having experts monitor, maintain, and support a company’s database 24/7 is no longer optional – whether that’s done internally or not.

Often, businesses simply cannot find the operational expertise required to deliver an always-on database environment. Or, there is always the risk of not being able to retain employed experts and their ingrained knowledge operating the database a company has decided on.

It is important to note that in the majority of DBaaS cases, companies are not looking to completely outsource and replace the management and operations or their database. Rather, they are looking to have more confidence – or some increased level of assurance – that their operational teams are being supported by experts that have real hands-on experience with the technology that they have deployed.

• Q. Specific to Apache Cassandra, are there other circumstances that might drive a business to switch over to a hosted/managed database from previously handling DB operations internally?

A. There is always such as a high demand for IT personnel with deep Apache Cassandra expertise. We often see companies losing key staff and then deciding that it’s time to reduce their risks and have an external, dedicated organization provide the Cassandra expertise.

We are also, predictably, a port-of-call for some businesses that have been self-managing a cluster when they suddenly encounter problems like a significant outage or a severely degraded production performance. Cassandra can be tricky to manage and maintain effectively – businesses that have been self-managing for many years suddenly encounter problems when scaling or where required maintenance activities are not performed on a regular basis.

• Q. What types of use cases are particularly suited to Apache Cassandra as opposed to other databases?

A. Apache Cassandra, an open source distributed NoSQL database, tends to be a popular choice when a business requires an operational DB with unlimited scale and extremely high availability (thereby delivering maximizing reliability). Cassandra achieves this through a masterless architecture and data replication. It’s a serious database for global scale applications that can support bi-directional multi-datacenter replication and deployment.

Some of the big Cassandra use cases that we see relate to the Internet of Things and the capture and analysis of time series data (BlackBerry, for example, has an interesting use case around this type of deployment).  Another significant application and use of Cassandra relates to enhanced customer experience for applications through customization and personalization. This use case is used for everything from loyalty programs to gaming applications and streaming of video content, and is particularly well suited for Cassandra.

• Q. What avoidable issues do enterprises frequently encounter when managing their own databases, and what specific practices should these companies adopt?

A. There are two key areas that we see customers getting into trouble with Cassandra and their database infrastructure.

First off, there can be a lack of effective maintenance and infrastructure hygiene. Cassandra, in particular, requires ongoing preventative maintenance activities to be performed that ensure the continued, efficient operation of the cluster. If these regular maintenance and tuning efforts are not performed on a regular basis, then issues can and will arise – usually suddenly and without any real warning.

Secondly, we see major issues around trying to scale to meet sudden demand. This complication is somewhat related to the first, in that to be capable of scaling rapidly your database and infrastructure have to be maintained and well tuned. For example, say you have an application that goes viral overnight. You probably need to scale your database significantly (and quickly), and if your Cassandra cluster is not tuned and ready for it, you’re likely in for a rocky time.

The bottom line is that you don’t get the amazing qualities of unlimited scale, extreme reliability, and high availability without some form of technical debt. That debt comes in the form of complexity with Cassandra. The key to avoiding issues is continued maintenance, monitoring, and scaling with care.

• Q. What should enterprises look for when evaluating a DBaaS strategy and providers?

A. Businesses should be looking for real and demonstrated operational experience and genuine expertise concentrated in the specific database technologies that they are deploying (whether Cassandra or otherwise). Any provider of note must have developed a deep knowledge base around the core technologies that they are supporting. Having a provider with a wide and varied set of technologies is more likely to produce a support-base that is broad but not necessarily deep. You really need technical capability depth when it comes to core critical infrastructure components. And, specifically when it comes to managing and supporting a database for production environments, you want to have confidence in the provider and an intimate understanding of the SLA.

• Q. What are Instaclustr’s specialties as a DBaaS provider?

A. I believe we differentiate through operational expertise, platform reliability, and support. Instaclustr’s team is built out on having deep technical expertise with Apache Cassandra, Spark, and core open source technologies that we regularly work with. It’s expertise that is backed by almost 10m node hours of operational experience with Cassandra. At this point, we’ve seen it all – through the good, bad, and ugly of deployments.

The ability to deliver a reliable service that we’ve engineered ourselves – in our managed services platform – is also important to us. This allows Instaclustr to automate important provisioning, maintenance, and monitoring activities, and also provides a ready-controlled interface tool for our technical operators to effectively engage with any managed cluster.

Lastly, I think a lot of it comes down to trust. Databases are the lifeblood to many, many businesses, so reliability and support is key in this industry. To that end, independent validation against SOC2 and other industry standards is always core to our strategy.

• Q. Can you share an example of an interesting customer use case that Instaclustr has been involved with?

A. One of the particularly unique startups that we are currently working with and supporting through free managed Cassandra infrastructure and services is a company from Kenya, AfricaBeat. Its aim is to promote Kenyan and African music and to provide an international stage for a number of creative artists in the region. The application will offer personalized recommendations for its user base and create local chart on the most popular tracks. The founders are so passionate about Kenyan and African music it is impossible not to be excited for them and their drive to continually improve the reach and quality of local music. And it just so happens that Cassandra has been the perfect database backbone to make this happen.

About the Author

Peter Nichol is the CEO of Instaclustr, which provides enterprise-grade, fully managed open source data infrastructure in the cloud. Nichol has been in high tech for 30 years, starting in the UK in 1986 when commercial software was just taking off. Since then he has lived on four continents and has held sales, marketing, and general management positions across multiple industries.