Every DBA has at some point asked: “What the heck happened to my database?” In this blog, we’ll review some of the more common scenarios where some little detail slips through the proverbial crack and that often results in some kind of database anomaly which momentarily raises tensions and blood pressures.

Supposedly nothing is new or has changed, yet without any obvious reason, database performance has suddenly noticeably slowed. Some possible causes and effects to look for:

  1. Database size has grown and crossed some threshold at which performance characteristics have changed either due to physical design limitation, query optimizer weakness, database statistics inadequacy, hardware resource bottlenecks, and numerous other potential issues caused simply by an increase in size reaching some inflection point. While this may appeal as the candidate of choice, it’s simply not often the sole authentic reason – but it can happen.
  2. Database object and/or metadata health have been become compromised by some unplanned and infrequently occurring boundary condition resulting in an unbalanced state. For example, an index may become unbalanced, statistics may become stale, a partition expansion might be problematic, and many other conceivably looming glitches exist to make life interesting. For most seasoned DBAs these types of issues are not that hard to spot and fairly easy to correct.
  3. Unenvisioned skews or imbalances in database workloads might occur whereby normal tasks and operations are radically impacted in an unforeseen manner. For example, at quarter end business users may submit too many reports all at once, or more likely some scheduled processes or jobs may overlap their execution in a manner that overly stresses the hardware and software resources. I can state from experience that this last issue was the most common issue in the production databases that I managed. As a DBA I did more job schedule balancing than I could have ever imagined.

This last issue has one more wrinkle to be aware of, virtualization. Now the DBA might not be able to as easily identify resource issues due to the shared nature of the resources adding an additional layer of complexity. Moreover, the virtualization infrastructure might further complicate proper forensic analysis. Assume that your database server virtual machine (VM) can be dynamically relocated to overcome some resource imbalances. Now imagine that a performance problem occurs this month, but that the dynamic VM relocation masks the impact of the issue. Thus, when the problem becomes perceptible the first question might well be when did it really first occur? It may have been a problem for weeks or months.

Now let’s move that same database server to the cloud.  Not only do you have the same virtualization related concerns about problems potentially being masked, but now there could be questions about whether we have purchased the correctly sized cloud components from the ever-growing catalog of deployment options available. Moreover, are we on shared, elastic disk storage when we should have maybe more properly chosen dedicated SSD or NVMe storage? What about network latency, are we split across regions or availability zones for an application whose nature requires something different? One could argue that one of the most useful artificial intelligence or machine learning applications the market needs is a cloud monitor, governor and advice tool. There simply aren’t enough knowledgeable experts on proper cloud provisioning. Most people are essentially stumbling their way through given that it’s so easy to scale up when wrong – it just costs more money.

Don’t get me wrong, I love the cloud. I am just smart enough to know that I don’t know all the answers and there are far fewer people who do at the current time than we really need. Not even the big cloud providers can hire enough people fast enough and train them to fill this gap. Therefore, for the next few years we are probably in for some interesting times. I think Robert Kennedy’s quote most apropos:

Like it or not, we live in interesting times. They are times of danger and uncertainty, but they are also the most creative of any time in the history of mankind. And everyone here will ultimately be judged – will ultimately judge himself – on the effort he has contributed to building a new world society and the extent to which his ideals and goals have shaped that effort.

About the Author

Bert Scalzo

Bert Scalzo is a guest-blogger for Quest and a renowned database expert, Oracle® ACE, author, database technology consultant, and formerly a member of Dell Software’s TOAD dev team. With three decades of Oracle® database experience to draw on, Bert’s webcasts garner high attendance and participation rates. His work history includes time at both Oracle Education and Oracle Consulting. Bert holds several Oracle Masters certifications and has an extensive academic background that includes a BS, MS and Ph.D. in computer science, as well as an MBA, and insurance industry designations. Bert is a highly sought-after speaker who has presented at numerous Oracle conferences and user groups, including OOW, ODTUG, IOUG, OAUG, RMOUG and many others. Bert enjoys sharing his vast knowledge on data modeling, database benchmarking, database tuning and optimization, "star schema" data warehouses, Linux® and VMware®. As a prolific writer, Bert has produced educational articles, papers and blogs for such well-respected publications as the Oracle Technology Network (OTN), Oracle Magazine, Oracle Informant, PC Week (eWeek), Dell Power Solutions Magazine, The LINUX Journal, LINUX.com, Oracle FAQ, Ask Toad and Toad World.

Start the discussion at forums.toadworld.com