NoSQL and automated databases – “DBAs can be the thought leaders”
As the end of the year approaches, we took some time to talk to John Pocknell, senior solutions product manager at Quest Software, about his vision of the future. We talk about database management, database automation, the rise of NoSQL and the evolving role of the database administrator or DBA.
JAXenter: Which issues do database management and monitoring currently have? In what ways can they profit from cloud technologies and automation?
John: One of the biggest challenges DBAs have is keeping up with the growing number and variety of platforms they have to manage. In fact, according to a survey carried out by Unisphere Research, more than two thirds (70 per cent) of DBAs are now managing more than 11 or more databases, and incredibly, 10 per cent were found to actually be managing more than 100! So, all this complexity combined with the astronomic growth of, and the need to protect data means DBAs are spending a disproportionate time dealing with tasks related to “keeping the lights on” and less time thinking more strategically. Automation, when used properly, can save DBAs many hours of time – according to many estimates, as much of 80% of their daily/weekly routine tasks could be automated.
Migration of databases to the cloud is one way to reduce the administration workload, but there is a trade-off. As a DBA, do you go for IaaS or DBaaS? With DBaaS, the cloud services provider manages your databases – backups, patching, security, etc, and you have some database monitoring capability for that specific database, but the extent to which you can manage other things like performance tuning, etc. are restricted. Unlike IaaS, where you manage the database, therefore the administrative workload is the same as on-premises, although the business costs (CapEx) will be lower.
An effective monitoring solution that works across all these different databases and environments is a must if you want to lower the administrative footprint. Added complexity means that your MTTR, in the event of a performance issue, will suffer without an effective way to detect, diagnose and resolve problems in a timely manner.
JAXenter: Aside from the positive effects, do you see any downsides in using cloud technologies for database administration?
One reason why we won’t see the role of the DBA go away with autonomous databases is because of the importance they play in ensuring the safety of data.
John: The downsides of moving databases to the cloud depend on the type of cloud service the company subscribes to. With DBaaS, the DBA stands the risk of a database being taken down at any time as the service provider performs a backup or applies a security patch (or worse still, the cloud provider suffers an outage). Also, the level of service the company chooses ultimately dictates the monthly subscription cost and many companies over-subscribe because they didn’t effectively right size the service ahead of time. It’s really important to choose the service tier that is most appropriate for acceptable application performance and that is sometimes a challenge because cloud database behave differently than in a data center.
Another ongoing concern is security and the protection of personal or sensitive data. Having personal data in the cloud does not obviate the company’s responsibility to identify and protect it in order to be compliant with data privacy regulations like GDPR and CCPA.
Vendor lock-in is another concern and there is a growing adoption of multi-cloud strategies where companies may choose different cloud providers according to application requirements, cost and risk balance, regional dependencies (e.g. EU provider if having to comply with GDPR).
SEE ALSO: Securing containers throughout the entire build-ship-run lifecycle means shifting left and right
JAXenter: In your opinion, what impact will autonomous databases have on DBA in the short and long term?
John: Some believe that autonomous databases will impact DBAs to the point that they’ll be out of a job. While I don’t agree with this, I do believe that the role of the DBA will certainly shift as AI and adaptive machine learning enable organizations to automatically self-patch, self-tune, detect anomalies, and optimize indexes much more quickly and efficiently than manual hands-on processing. DBA’s, now free from many of their mundane tasks, will have to take on more strategic roles in developing new business initiatives and focusing on finding new areas where they can provide value, such as DevOps.
One reason why we won’t see the role of the DBA go away with autonomous databases is because of the importance they play in ensuring the safety of data. DBAs are responsible for implementing data privacy regulations, ensuring compliance, and as DevSecOps grows, they’ll begin to enable things like proper testing in production environments.
In fact, DBAs should consider reinventing themselves as “data administrators”, since the value of data to the company is huge and the DBA is a key stakeholder who can help drive the business forward. They can be the thought leaders in what database technologies are most appropriate for their applications and help shape company strategy.
JAXenter: NoSQL has been on the rise as well. Do you believe we will keep heading in that direction over the next year?
The other benefit of using NoSQL that simplifies the DBA’s job is scalability.
John: Yes, in fact, 75 percent of companies are already using both SQL and NoSQL databases with MongoDB and Cassandra being among the most popular, according to the cloud database trends report from DeveloperWeek 2019.
NoSQL hasn’t seen a huge amount of movement in recent years, but I believe we’ll see it pick up more next year, especially as people move towards fresher and newer data needs. While relational databases are good for traditional workloads like OLTP applications and business analytics (OLAP), for more complex OLTP workloads that include low-latency applications, NoSQL is better (versatility, agility, scalability). Ultimately, it’s a matter of getting the right database to suit the workloads of the organization, especially with the variety of structured and unstructured data in use. I also think we’ll also see adoption of cloud NoSQL databases (such as Amazon’s DynamoDB and Google’s Cloud Datastore).
SEE ALSO: Software development trends for 2020
JAXenter: Which implications will the increased use of NoSQL have, e.g. regarding the way DevOps and DBA interact?
John: Depending on the type of application, NoSQL databases offer some advantages in a DevOps CI/CD pipeline that can hinder the use of relational databases. Synchronizing relational database changes with an object-oriented application can be a challenge (called impedance mismatch) and requires application developers to leverage Object Relational Mapping (ORM) classes in their application code to relate to the database schema structure and any changes that might occur (e.g. change a column, add a table). This is something DBAs and the application team must manage to ensure things don’t break.
With the growth of the use of NoSQL document databases such as MongoDB and CouchDB, this problem goes away since there is no schema in the traditional sense, meaning application developers can arrange their data (structured, semi-structured, unstructured) how they want and remove the impedance mismatch that ORM was designed to resolve.
The other benefit of using NoSQL that simplifies the DBA’s job is scalability. NoSQL databases leverage commodity hardware and therefore are far easier to scale out compared with relational databases.
Thanks very much!