The Reality of Database as a Service (DBaaS)

Posted By:  Brian Feeny, Director of Engineering, South Florida, Presidio
Posted Date: 

There has been a proliferation in cloud offered services over the last few years.  Almost every vendor now has a cloud like offering or is selling their technology or software as a service. These services mean to simplify the costs and effort needed to deploy and maintain IT services. One of the latest offerings is Database as a Service.

Data is possibly the most important asset a business has.

Databases can be found everywhere in an enterprise. There are usually hundreds of databases and many have thousands. Just about every operational data store or reporting tool requires a database of some sort. Because departments within the enterprise have started to become autonomous from IT, you end up with a sprawl of databases, sometimes being managed by those with very little database knowledge or time to manage it.

Enter the appeal of a DBaaS. Companies have made their database products available in the cloud. This simplifies the startup and capitalization of database services, as well as the ongoing maintenance and support. Many of these services are “pay as you go” and have rather attractive price points. The amount of responsibility undertaken by the DBaaS provider generally includes:

  • Installation of the database server(s) and software
  • Patches and upgrades
  • Hardware lifecycle management
  • Monitoring and Troubleshooting
  • Security
  • Backup and Recovery

Some of these services are included as part of a standard offering, for example the installation and maintenance of the server and software. Other parts, such as backup and recovery, may be additional charges with different service levels.

What’s not to like?

With so many responsibilities being delegated to the DBaaS provider, combined with a simple “pay as you go” model, what’s not to like? Well, as the saying goes, “There’s no free lunch.”, and like everything in IT, making a decision to go to the cloud can have its tradeoffs.

Here are a few reasons that anyone considering DBaaS should proceed with caution.

Availability and Uptime - Moving your data into the cloud can add increased complexity and potential for failure between you and your data. WAN links failure and service provider outages are just two of the things that are outside of your control when you’re in the cloud. Because the infrastructure is shared, it’s possible that a problem with one part of the overall system may affect your experience.

Latency and Response Time - Some users expect very fast responses from their databases. This is common with operational data stores or OLTP databases. Because there are numerous amounts of shared infrastructure supporting the database, including the Internet itself, response times can be very inconsistent. Reporting databases or OLAP databases can use massive amounts of data. Loading data in or out of a database, known as ETL, can take much longer when dealing with a database in the cloud.

Security - Today’s enterprises are data driven. The data can be some of the most important information a business has. It can also be the Achilles heel that takes a business down, as consumers are increasingly sensitive to breaches in their information. When using a large cloud provider, you have a very large attack surface. Even if an attacker is not interested in your specific data, they may have a goal of breaching the larger provider, which can spell trouble for your data and your goodwill.

Data Governance - Another important consideration for enterprises is who has access to the data, and what access do they have. When storing data on a cloud provider it’s possible there are people that potentially have access to the data outside of your organization. Every organization has their own level of comfort depending on the type of data being stored.

These are just a few of the considerations that must be looked at when thinking of moving your data into a DBaaS. The list continues but these are the top four considerations.

With so many concerns, is there any data good for DBaaS?

There are conveniences and risks involved when looking to move data into a DBaaS. Only your organization stakeholders can make the decision as to whether or not they are comfortable with the tradeoffs. Some potential areas that are good candidates for initial use cases for DBaaS include:

Proof of Concepts - The ease of getting up and running make DBaaS a good candidate to evaluate various database vendors. You have an environment where you can move some of your data into and try out your tools to see if the database has good interoperability with your workflow. Organizations should be cautioned in doing critical performance testing of different software packages in cloud based environments compared to on-premise environments because there are many factors in the equation besides just the software. Use the cloud to test drive the database and if you truly wish to make a comparison, evaluate both products in the cloud or on-premise.

Non-sensitive data - Not all data is equally valuable. For data where the risk is low, and the data are potentially masked, DBaaS can be a candidate.

Test, sandbox and development environments - The ideal test environment mirrors the production environment. However some organizations in an attempt to be more agile, can utilize DBaaS environments to simplify their infrastructure footprint and decrease database server sprawl.

DBaaS Alternatives

Many organizations investigating DBaaS want to simplify their infrastructure, reduce database server sprawl and reduce cost. It is possible to address these issues by simply removing the numerous soloed databases and move to more of a centralized environment known as an Enterprise Data Warehouse (EDW). A modern EDW environment allows organizations to consolidate numerous databases and have better control of their data through economies of scale. The EDW allows organizations to store both structured and unstructured data. Infrastructure is greatly reduced as is software, support and maintenance costs. Best of all, the data remains on-premise and within the control of the enterprise reducing much risk and increasing overall service levels.

Extending the Private Cloud

Before any DBaaS is implemented, an organization should evaluate what the cost benefits are of doing this with a private cloud or other creative solutions like Capacity on Demand. Long term costs of ownership analysis has shown that leveraging a private cloud and then extending it is more cost effective than services offering via public cloud providers. In additions most of the concerns outlined in the document are addressed.

Conclusion

Data is possibly the most important asset a business has. Businesses today are trying hard to deal with the explosion of data and leverage it to their advantage. Although DBaaS can be appealing, and it is appropriate for some use cases, it is important to carefully weight the tradeoffs and take the course of action that allows your business to operationalize and monetize its data to its full potential.