ScaleDB and Oracle RAC are both clustered databases that use a shared-disk architecture. As I have mentioned previously, they both actually share data via a shared cache, so it might be more appropriate to call them shared-cache databases.

Whether it is called shared-disk or shared-cache, these databases must orchestrate the sharing of a single set of data amongst multiple nodes. This introduces two challenges: the physical sharing of the data and the logical sharing of the data.

Physical Sharing:
Raw storage is meant to work on a 1:1 basis with a single server. In order to share that data amongst multiple servers, you need either a Network File System (NFS), which shares whole files, or a Cluster File System (CFS), which shares data blocks.

Logical Sharing:
This is specific to …

Last night I was the invited guest at the SF MySQL Meetup. In my presentation “Successful MySQL Scalability” I talked about a set of principles to ensure appropriate system architecture, data availability and best practices to build an ideal solution for your business. The presentation was also live streamed and is available online.

Successful MySQL Scalability

In this latest release I’ve added a basic MySQL connection pool to the benchmarking script which improves the method in which connections to MySQL are handled and reused. In addition, there have been some optimizations made to the thread handler functions for better debug reporting. Download the latest release now and see how your MySQL server performs against the rest of the community! Download here: download page.

Few days ago I was working on a case where we needed to modify a lot of data before pushing it to sphinx – MySQL did not have a function to do the thing so I thought I’ll write MySQL Stored Function and we’ll be good to go. It worked! But not so well really – building the index, which was taking 10 minutes, was now taking 16 minutes. Then we added another MySQL function for different set of attributes and indexing speed went from 16 minutes to 26 minutes. I knew using UDF would be faster, but I had no idea how much. Have you ever wondered?

So what were the modifications we needed? It was couple very simple things – (1) two varchar columns needed leading nonalpha characters trimmed, so “123 ^&* …

In my work on group commit for MariaDB, I have the following situation:

A group of threads are going to participate in group commit. This means that one of the threads, called the group leader, will run an fsync() for all of them, while the other threads wait. Once the group leader is done, it needs to wake up all of the other threads.

The obvious way to do this is to have the group leader call pthread_cond_broadcast() on a condition that the other threads are waiting for with pthread_cond_wait():

  bool wakeup= false;
  pthread_cond_t wakeup_cond;
  pthread_mutex_t wakeup_mutex

Waiter:

  pthread_mutex_lock(&wakeup_mutex);
  while (!wakeup)
    pthread_cond_wait(&wakeup_cond, &wakeup_mutex);
  pthread_mutex_unlock(&wakeup_mutex);
  // Continue processing after group commit …

In my work on group commit for MariaDB, I have the following situation:

A group of threads are going to participate in group commit. This means that one of the threads, called the group leader, will run an fsync() for all of them, while the other threads wait. Once the group leader is done, it needs to wake up all of the other threads.

The obvious way to do this is to have the group leader call pthread_cond_broadcast() on a condition that the other threads are waiting for with pthread_cond_wait():

  bool wakeup= false;
  pthread_cond_t wakeup_cond;
  pthread_mutex_t wakeup_mutex

Waiter:

  pthread_mutex_lock(&wakeup_mutex);
  while (!wakeup)
    pthread_cond_wait(&wakeup_cond, &wakeup_mutex);
  pthread_mutex_unlock(&wakeup_mutex);
  // Continue processing after group commit …

I get a number of question about contentions/"stuck in..". So here comes some explanation to:

• Contention
• Thread Stuck in
• What you can do about it
  

In 99% of the cases the contentions written out in the out file of the data nodes (ndb_X_out.log) is nothing to pay attention to.

sendbufferpool waiting for lock, contentions: 6000 spins: 489200
sendbufferpool waiting for lock, contentions: 6200 spins: 494721

Each spin is read from the L1 cache (4 cycles on a Nehalem (3.2GHz), so about a nanosecond).
1 spin = 1.25E-09 seconds (1.25ns)

In the above we have:
(494721-489200)/(6200-6000)= 27 spins/contention
Time spent on a contention=27 x 1.25E-09=3.375E-08 seconds (0.03375 us)

So we don't have a problem..

Another example (here is a lock guarding a job buffer (JBA = JobBuffer A, in …

The development cycle is moving right along for the community’s newest MySQL benchmarking script. I’m pleased to announce that we now officially support FreeBSD (version 8.1 tested) so go ahead and download now and test your FreeBSD, Linux, or OSX MySQL server! Click here for the download.

Courtesy of Darren Cassar and some generous coding this weekend, we’re going to be releasing a auto-installer / updater for the application which you can use to automate that part of the process. Stay tuned for information on that release.

We had a very successful weekend of Planet.mysql users submitting their database statistics so I’ve pushed some code into a new release today so that everyone can benefit from some new features. The biggest change is to the threading logic. Previously the benchmarking script was serializing MySQL operations and only making use of a secondary thread (not the invoking thread) to query the database. Now you have the option of running with “–threads=x” to make use of your multi-core server. A good example of this improvement was on my Macbook Pro; before the threading change it was inserting ~700/sec, after the code change I tried –threads=4 and saw an improvement to ~900/sec. Rather significant.

Download the new script now and see how your server compares to the ones in the …

This has really been a long debate as to which approach is more performance orientated, normalized databases or denormalized databases. So this article is a step on my part to figure out the right strategy, because neither one of these approaches can be rejected outright. I will start of by discussing the pros and cons of both the approaches. Pros and Cons of a Normalized database design. Normalized databases fair very well under conditions where the applications are write-intensive and the write-load is more than the read-load. This is because of the following reasons: Normalized tables are usually smaller and...