Database Server Performance Tuning

Improving database performance is essential, as it increases the system’s ability to process information per unit of time, reduces costs through lower hardware usage (fewer servers or less cloud time), and provides a competitive edge by delivering faster and more comprehensive information, reports, and dashboards—ultimately becoming a strategic management tool for business units.

Database Server Performance Tuning is a premium service offering fine-tuned optimization. While many competitors attempt to resolve performance issues by focusing solely on server parameter settings, we take a different approach. Our team performs a deep dive into the entire database environment, thoroughly analyzing all contributing factors that can either enhance or hinder overall performance.

When to Hire a Database Performance Tuning Service:

A database that previously delivered satisfactory performance has started to show signs of degradation.
After an update (of the RDBMS itself, the application, data modeling, metadata, or SQL code), a drop in performance was noticed.
Data consumers express dissatisfaction with the delivery times of reports and queries.
The SLAs for screens, applications, reports, and integrations are compromised.
Even when everything seems to be working fine, but you need to be sure you’re getting the best possible performance.

Check out

News portal

Reducing the time it took to display news helped prevent subscriber churn. With improved scalability, it became possible to grow the subscriber base. Increased performance and available resources enabled the creation of more personalized banners tailored to each user’s profile, boosting advertising monetization rates.

Online Games

Through consulting focused on parameter optimization and horizontal scalability, we increased the number of concurrent players by 5x, resulting in gains in subscriptions and advertising revenue.

E-commerce

Reducing navigation time between categories, products, and the checkout process increased the number of visitors and, consequently, boosted sales.

WMS and Tax

The performance improvement enabled faster waves of goods picking and reduced the time required for route and load planning, resolving the recurring issues with order backlog.

Financial Services

Deliver complete and secure information to your clients quickly in order to accelerate credit decisions. As the database and number of requests grew, the system nearly came to a halt. With our performance tuning and high-volume support service, we helped our client achieve their goals—and those of their clients.

Insurance

In a SOA (Service-Oriented Architecture) environment, it is essential that data consumption services are fast. Through a performance consulting project involving the rewriting of procedures, queries, and server parameter tuning, we were able to reduce response times by up to 75% for critical services related to screening and proposals.

How Our Service Works

1. Consumption and Utilization

Consumption and Usage Metrics

The first step is an analysis of the entire load and consumption of your database. Using our exclusive database monitoring tool, we will collect usage statistics to better understand how and how much the database is consumed: transactions per second, number and types of locks, concurrency, form and amount of usage, consumption peaks, execution plans, hardware and operating system behavior.

2. Hardware

Demand Adherence

There has always been a maxim that databases should only run on physical machines. With new virtualization technologies, this is no longer an absolute truth. However, it is essential to understand whether the hardware and/or virtual machine is properly configured and whether the available resources are sufficient to meet the demands placed on the database. Incorrect capacity or configuration of hardware and virtual machines can account for a significant portion of poor performance. But never believe that “just adding more hardware will solve everything” — we don’t believe in that. We prefer to solve database performance issues with intelligence, not with cash flow.

3. Operating System

Operating System Tuning

The Operating System is often one of the most overlooked aspects when setting up a new database environment. Typically, there’s an “established” OS deployment standard that is followed without considering the specific requirements of the chosen RDBMS.

Our approach is to align and adapt this standard to the needs of the RDBMS. We address contingencies and tune the OS through configuration adjustments, checkpoint strategies, caching methods, disk presentation, file system types, and their block settings.

4. Parameter Configuration

Optimization = Hardware + Operating System + Demand

Depending on the database (Oracle, MySQL, SQL Server, DB2, PostgreSQL, etc.), parameter tuning can be the most critical part of a performance tuning effort.
However, it’s important to acknowledge that there’s a limit to what can actually be achieved by simply “tightening the screws.” There’s no more magic in that hat. Parameter tuning is indeed important, but it’s not the Holy Grail of Performance. There’s no magical command like ALTER PERFORMANCE = NITRO.

That said, a perfectly orchestrated configuration of parameters can contribute significantly to performance improvement. It’s about balance, context, and a deep understanding of how each parameter interacts with the system’s workload.

5. Parallelism Strategy

Scalability and Performance

How much and what should be parallelized? Cores, Threads, NUMA… Modern hardware architectures allow us to extract the full potential of CPUs (chips) and leverage the use and mapping of large amounts of memory (RAM). However, this requires a systemic understanding of the whole: hardware + operating system + database configuration.

Under-parallelization leads to underutilization of hardware potential, limiting performance. Conversely, excessive or poorly strategized parallelization can exhaust hardware resources, significantly increasing conflicts, contention, locking, and MUTEX issues — which in some cases drastically reduce performance.

That’s why we emphasize the word “strategy”. The cores and threads are there — knowing how much to use is just as important as properly tuning the RDBMS and OS.

6. Indexing Strategy

Selectivity and Cardinality

Are the indexes built in the most efficient way to serve your queries?
Are there rarely or never used indexes that are impacting write performance? Everyone has heard: “Just create an index and the query time will drop.” Or even better: “We need to drop some indexes to improve write performance.” But what about the time spent maintaining those indexes?

Databases have a life of their own! Cardinality and selectivity change, and with that, good indexes can become bad—and sometimes, bad ones may become good. Again, the key word is strategy. We will analyze your indexes, how they were designed, and whether they are serving your most critical and demanding queries.

7. Data Modeling

Normalization Level and Relationships

A brief study of data modeling and how it can impact overall database performance. From an academic standpoint, we believe that the success or failure of an application begins with the modeling of its database. Is the chosen database system the most suitable for the intended application? Would a normalized structure perform better than a denormalized one? Should the data be stored in a two-dimensional model (columns x rows), documents, column families, or graphs? Am I certain about all the questions I want to “ask” my database? Have I mapped out the best ways to search for those answers?

In reality, taking proper care of data modeling is an essential part of the software development process. Once the system is in production, making changes is as difficult as fixing an airplane engine mid-flight. Refactoring the database could mean hours of rewriting application source code, which involves significant human and financial resources.

Therefore, while we don’t delve deeply into this foundational area within the scope of our performance tuning service (we have a specific service for that), we do focus on aspects of modeling that may negatively impact performance. These include excessive normalization, inappropriate data types, staging tables, temporary tables, summary tables, and other factors that can be effectively adjusted in alignment with the application’s needs.

8. Offensive Queries

Execution Plan and Cost

Identification of the most resource-intensive and/or low-performing queries, along with recommendations for best practices to be adopted in their refactoring. Additionally, a specialized service can be contracted for SQL code refactoring (queries, triggers, procedures, functions, jobs, events).

Contact us and get your questions answered

Do you have questions about our services or need technical assistance? Fill out the form below and our team will get in touch with you as soon as possible.