Mark BurgessThe Real Cost of Poor Performance
Optimising and tuning Oracle systems for optimal performance is a well-covered topic. There are a range of tools and methodologies to help support engineers resolve system performance issues, quickly and correctly.
Why, with the advance of features in the software along with tools and techniques to fix performance problems do we still encounter issues?
Why is it that with the significant investments made in Oracle technology, many organisations are still constrained by the performance of their database systems?
The answer is, we often don’t understand the underlying business cost.
The Cost to Indirect Stakeholders
The vast majority of critical applications deployed on Oracle technology support the job of an indirect stakeholder of the system:
- Call centre teams who require their CRM application screens to open as fast as possible.
- Payroll teams who are required to process employee payrolls on time and on schedule.
- Executive management who require accurate and timely dashboard reporting.
- Purchasing customers you might easily lose due to a slow website.
Often, indirect stakeholders will accept the level of performance that they are provided.
“It’s always been that slow”
“We’ve given up asking for improvements”
”The technical teams say that it can’t be done”
Sound familiar?
We need to ask if this is good enough.
Is it good enough that an executive dashboard takes 30 seconds to return a report when it could be 2 seconds?
Is it good enough that the payroll team need to put in overtime every pay cycle when it could be done in minutes?
Is it good enough that customers are lost due to slow loading payment pages?
The indirect cost of poor performance is difficult to measure in financial terms but it is real.
Performance “cost” is typically discussed in technical terms and direct financial cost to remediate a system measurement.
IT investment cost only covers a part of the equation. The intangible stakeholder cost also needs to be considered as part of the decision to remediate system performance issues.
In order to invest appropriately, we as technologists, need to understand cost from the indirect stakeholder perspective.
The benefit of this approach is twofold:
- It helps to identify and fix the performance problems impacting the organisation.
- It helps to secure budgets to resolve performance issues.
System Performance Must Be Timely to be Useful
Currently (and most likely for a very long time before AI makes the human brain obsolete) analytic reporting is solely a tool for humans to interpret and use.
But it doesn’t matter how pretty the analytics look, they are useless if they are not delivered in a timely manner.
If reports are not generated at a cadence to match decision making, the hidden cost is organisational success. Or, put another way, users don’t use the analytics and the incremental value they can provide to the organisation is wasted.
Typical analytic rollouts occur incrementally, with ongoing refinement of the platform. As the analytic capability evolves, users ask more complex questions from the platform. Read more about Operational Design Requirements.
However, if system performance doesn’t keep up, and the platform is unable to provide timely answers, those insights have no value.
On the other hand, organisations that can continue to keep pace with user performance requirements can develop a huge advantage. Decision makers are provided with timely insights, aiding decision making and driving business performance.
Long-Term performance tuning of the analysis system, allows the platform to keep up with user needs and deliver incremental business performance.
The Hidden Cost Of Oracle Database Environment Consolidation
Driving lower costs by consolidating Oracle database environments into a service-based model has been an ongoing theme for a number of years.
However, savings from consolidating infrastructure, license and operations procedures can be quickly eroded when workloads are not optimised.
In the best case, additional infrastructure (hardware and software) and cost, is required to support sub-optimal workloads.
But in the worst case scenario, consolidation of sub-optimal workloads actually amplifies poorly performing workloads, reducing overall performance even further.
Stakeholder experience from these poorly performing systems falls off a cliff and the original business case for consolidation is at risk. Missed cost savings and efficiencies from database consolidation then have a direct impact on future business investment confidence.
Ensuring systems are appropriately tuned before consolidation prevents amplification of performance issues and secures long-term investment opportunities.
Generic Application Schema Design
Many application vendors “abstract” the application code from the database platform. This allows the customer to deploy the application to the customers preferred database platform.
The flexibility this deployment grants can come with significant hidden cost to Oracle performance as data volumes increase. Inefficient SQL query plans, significantly increased resource consumption and corresponding infrastructure and license costs are all silent costs of “generic” application design.
Be aware that typically problems that are a result of “generic” application do not present themselves until the system has been in operation for some time – and can become difficult and expensive to fix.
The performance cost of database portability can be significant – increased license costs to accommodate inefficient workloads, generally poor response time on critical functionality, expensive and inefficient solutions to work around database specific issues.
Putting Poorly Performing Systems in the Cloud Costs More Not Less
Deploying sub-optimal workloads to a cloud environment does not remove the need to minimise waste workloads in the system. In fact, it can make the total cost of supporting these workloads significantly worse.
Sub-optimal, on-premise, workloads consume resources from a platform that is typically accounted for as a capital expense.
For example, say, your poorly performing workload consumes 60% of your $1m investment. Then, from an accounting point of view, you are consuming 60% of the value of the depreciating asset over time. Because the value of your asset decreases each year, it costs less to support the wasteful workload.
On the other hand, if you migrate that workload to a cloud environment, you will be paying 60% of your subscription to support that sub-optimal workload – this year, next year, every year.
Over 3-5 years in an opex model (typical for cloud service subscriptions) that poorly performing workload is costing substantially more than it would on-premise.
So, before migrating database workloads to a cloud environment, it is vital to ensure the system is tuned and waste workloads are minimised. Not doing so can have a significant impact on the cloud business case.
Conclusion
The impact of poor Oracle performance optimisations is substantial but is often only revealed indirectly outside the technical jargon.
Additionally, some apparent shortcuts that imply Oracle performance optimisation is optional, can exacerbate the performance and costs issues that they look to resolve.
Start eliminating waste workloads with our proven diagnostic process.
