Press "Enter" to skip to content

Category: Performance Tuning

Tips for Decreasing the Impact of Rebalancing in Kafka Streams

Published 2021-07-28 by Kevin Feasel

Vasyl Sarzhynskyi has some techniques to make rebalancing in Kafka less of a big deal:

Kafka Rebalance happens when a new consumer is either added (joined) into the consumer group or removed (left). It becomes dramatic during application service deployment rollout, as multiple instances restarted at the same time, and rebalance latency significantly increasing. During rebalance, consumers stop processing messages for some period of time, and, as a result, processing of events from a topic happens with some delay. Some business cases could tolerate rebalancing, meanwhile, others require real-time event processing and it’s painful to have delays in more than a few seconds. Here we will try to figure out how to decrease rebalance for Kafka-Streams clients (even though some tips will be useful for other Kafka consumer clients as well).

Read on for an example of the problem, as well as several practical tips for mitigating the issue.

Leave a Comment

Minimizing PAGELATCH_EX Waits in SQL Server 2019

Published 2021-07-27 by Kevin Feasel

Aaron Bertrand has some advice for us:

There is a common problem in SQL Server where contention when writing to a single page can lead to excessive waiting. This waiting comes in the form of the wait type PAGELATCH_EX, and often happens when the clustering key is a monotonically increasing value – like an IDENTITY or date/time column. Pedro Lopes has highlighted some non-trivial workarounds in his post, “PAGELATCH_EX waits and heavy inserts,” and there is some official guidance in the Microsoft Docs article, “Resolve last-page insert PAGELATCH_EX contention in SQL Server.” But are there any ways to address this issue without intrusive changes?

Read on to see what Aaron has in mind.

Leave a Comment

Performance Issues with AT TIME ZONE

Published 2021-07-23 by Kevin Feasel

Jonathan Kehayias lays out a warning:

This is not the type of blog post that I enjoy writing. First let me start off by saying this is not a bash against Microsoft, the SQL Server team, or anything other than an informative post to let people know that there is a potential performance limitation in a very useful feature of SQL Server 2016 and higher. The AT TIME ZONE syntax was added in SQL Server 2016 to handle changing datetime values from one time zone to the offset of a different time zone using string names for the time zone. This is a great feature that simplifies converting datetime values but there is an unfortunate draw back to the implementation; it relies on the time zones that are stored in the Windows Registry and therefore has to make calls out to Windows which unfortunately occurs row-by-row for large result sets when used inside of a query.

Read on for more detail, as well as the way that Jonathan fixed the performance issue.

Leave a Comment

The Benefits of a Cache Layer

Published 2021-07-15 by Kevin Feasel

Monica Rathbun espouses upon the utility of Azure Cache:

One of the biggest impacts on resource consumption for Azure SQL DB are repeated data pulls by the application layer. No matter how fast those queries execute calling the same procedure or issuing the same SQL statements hundreds, thousands, or million times a day can wreak havoc on database performance. Death by a thousand cuts can easily bring a system to its knees. Sometimes it’s hard for DBAs to troubleshoot these actively as the execution of the statements happens so quickly they don’t even show in tools like sp_whoisactive. It’s not until you begin to dive into things like Query Performance Insights or Query Store that you start to see the real issue.

Check it out. And if you want to get into implementation, I’ve found the Cache-Aside design pattern to be useful. Bowen Li has a rundown of several caching patterns as well.

Leave a Comment

High-Performance ETL via Buffer Table

Published 2021-07-15 by Kevin Feasel

Daniel Hutmacher needs things to zoom:

It’s almost like a myth – one that I’ve heard people talk about, but never actually seen myself. The “shock absorber” is a pretty clever data flow design pattern to ingest data where a regular ETL process would choke on the throughput or spikes. The idea is to use a buffer table to capture incoming data, and then run an asynchronous process that loads that data in batches from the buffer into its intended target table.

While I’ve seen whitepapers and blog posts mention the concept loosely along with claims of “7x or 10x performance”, none of them go into technical detail on how it’s done, so I decided to try my hand at it.

I’ve compiled my findings, along with some pre-baked framework code if you want to try building something yourself. Professional driver on closed roads. It’s gonna get pretty technical.

Combine that with Eitan Blumin’s post yesterday and you’d think it were buffer week.

This shock absorber pattern works well for warehouse loading, especially when you’re trickle-loading data into columnstore indexes and don’t want to have open rowgroups slowing everything down.

Leave a Comment

Buffering Events in SQL Server

Published 2021-07-14 by Kevin Feasel

Eitan Blumin has a technique to reduce expensive upserts:

Do you find yourself facing performance problems and long lock chains caused by very frequent INSERT, UPDATE, or DELETE statements being executed on a table? Check out this neat trick that could help you out and make all the difference in the world.

Okay, I admit that title ended up being quite long. But I wanted something that could be easily found in search engines by people facing similar problems.

I’ve done something similar, though without the partition switch and instead deleting batches into a temp table. This is a good example of something I like to say about scalable processes in T-SQL: many times, the most scalable technique involves a mental pivot (and sometimes a literal pivot, such as using tally tables to work with string data) of the straightforward answer.

1 Comment

Identifying Backpressure in Apache Flink

Published 2021-07-08 by Kevin Feasel

Piotr Nowojski explains an important concept in streaming (and ELT/ETL) products:

The backpressure topic was tackled from different angles over the last couple of years. However, when it comes to identifying and analyzing sources of backpressure, things have changed quite a bit in the recent Flink releases (especially with new additions to metrics and the web UI in Flink 1.13). This post will try to clarify some of these changes and go into more detail about how to track down the source of backpressure, but first…

Read on for the full story, including a review of the concept and its importance.

Comments closed

UI Patterns which Clash with Database Patterns

Published 2021-07-08 by Kevin Feasel

Michael J. Swart explains why we can’t have nice things:

I spend a large amount of time translating software requirements into schema and queries. These requirements are sometimes easy to implement but are often difficult. I want to talk about UI design choices that lead to data access patterns that are awkward to implement using SQL Server.

Read on for three such examples, including sorting, paging, and search.

Comments closed

Using DBCC INPUTBUFFER

Published 2021-07-08 by Kevin Feasel

Monica Rathbun shows us how to use DBCC INPUTBUFFER:

A command I like to use when performance tuning is DBCC INPUTBUFFER. If you have ever run sp_whoisactive or sp_who2 to find out what sessions are executing when CPU is high for instance this can be a real quick life saver. At times, for me, those two options do not return enough information for what I’m looking for which is the associated stored procedure or object. Using this little helper along with the session id can easily get you that information.

Let’s take a look.

Let’s.

Comments closed

Parallelizing R Code

Published 2021-07-02 by Kevin Feasel

Mira Celine Klein walks us through some of the basics of parallel code execution in R:

In many cases, your code fulfills multiple independent tasks, for example, if you do a simulation with five different parameter sets. The five processes don’t need to communicate with each other, and they don’t need any result from any other process. They could even be run simultaneously on five different computers… or processor cores. This is called parallelization. Modern desktop computers usually have 16 or more processor cores. To find out how many cores you have on your PC, use the function detectCores(). By default, R uses only one core, but this article tells you how to use multiple cores. If your simulation needs 20 hours to complete with one core, you may get your results within four hours thanks to parallelization!

Read on to see how you can accomplish this, but note that it is operating system-dependent.

Comments closed