Jun
30
2011
--

Percona Server 5.5.13-20.4 Stable Release

Percona is glad to announce the release of Percona Server 5.5.13-20.4 on July 1st, 2011 (Downloads are available here and from the Percona Software Repositories).

Based on MySQL 5.5.13, Percona Server Percona Server 5.5.13-20.4 is now the current stable release in the 5.5 series. All of Percona’s software is open-source and free, all the details of the release can be found in the 5.5.13-20.4 milestone at Launchpad.

Improvements

SHM Buffer Pool has been replaced with LRU Dump/Restore

The ”SHM” buffer pool patch, which provided the ability to use a shared memory segment for the buffer pool to enable faster server restarts, has been removed. Instead, we recommend using the LRU Dump/Restore patch which provides similar improvements in restart performance.

Replacement is due to SHM buffer pool both being very invasive and not widely used. Improved restart times are better provided by the much safer LRU D/R patch which has the advantage of also persisting across machine restarts.

The configuration variables for my.cnf have been kept for compatibility and warnings will be printed for the deprecated options (innodb_buffer_pool_shm_key and innodb_buffer_pool_shm_checksum) if used.

Instructions for disabling the SHM buffer pool can be found here.

Instructions on setting up LRU dump/restore can be found here.

Bug Fixes

  • On a high concurrency environment with compressed tables, users may experience crashes due to improper mutex handling inbuf_page_get_zip(). Bug Fix: #802348 (Y. Kinoshita).
  • XtraDB crashed when importing big tables (e.g. 350G) using the Expand Table Import feature due to a timeout. Bug Fix: #684829 (Y. Kinoshita).
  • Partitioning adaptive hash index may leave to a hangup of the server in some scenarios. Bug Fix: #791030 (Y. Kinoshita).
  • Statistics gathering for each record’s update. Bug #791092 (Y. Kinoshita)

Other Changes

  • Improvements and fixes on platform-specific distribution: #737947

For more information, please see the following links:

Jun
23
2011
--

FusionIO 320GB MLC random write performance

I have posted results for FusionIO 320GB MLC on our ssdperformanceblog.com blog.
To not duplicate content, there is link on original post:
http://www.ssdperformanceblog.com/2011/06/fusionio-320gb-mlc-random-write-performance/

Follow my twitter
for further results.

Jun
23
2011
--

Intel 320 SSD random write performance

I have posted results for Intel 320 SSD on our ssdperformanceblog.com blog.
To not duplicate content, there is link on original post:
http://www.ssdperformanceblog.com/2011/06/intel-320-ssd-random-write-performance/

Follow my twitter @VadimTk for further results.

Jun
23
2011
--

Dear Database Vendors

I have one wish for each database. Some of them are big wishes, so I thought I’ll better send them early. Only six month until Christmas, you know ;)

Oracle

” IS NULL? Are you serious? I know, it has always been that way, but does that mean it will stay so forever?

SQL Server

Could you please enable READ_COMMITTED_SNAPSHOT per default? Or, at least, make it a mandatory question during database creation? That could prevent so many locking problems.

PostgreSQL

Could you please implement index-only scans (aka “covering indexes”)? It’s a great performance feature, which I really miss in PostgreSQL.

MySQL

Could you please implement the hash join algorithm? So many applications suffer from poor join performance, just because MySQL doesn’t have hash joins.

From today’s perspective, it looks like my PostgreSQL wish is the only one that might come true. But I’ll keep hoping for the others as well.

Jun
23
2011
--

Dear Database Vendors

I have one wish for each database. Some of them are big wishes, so I thought I’ll better send them early. Only six month until Christmas, you know ;)

Oracle

” IS NULL? Are you serious? I know, it has always been that way, but does that mean it will stay so forever?

SQL Server

Could you please enable READ_COMMITTED_SNAPSHOT per default? Or, at least, make it a mandatory question during database creation? That could prevent so many locking problems.

PostgreSQL

Could you please implement index-only scans (aka “covering indexes”)? It’s a great performance feature, which I really miss in PostgreSQL.

MySQL

Could you please implement the hash join algorithm? So many applications suffer from poor join performance, just because MySQL doesn’t have hash joins.

From today’s perspective, it looks like my PostgreSQL wish is the only one that might come true. But I’ll keep hoping for the others as well.

Jun
22
2011
--

FusionIO 320GB MLC random write performance

I have posted results for FusionIO 320GB MLC on our ssdperformanceblog.com blog.
To not duplicate content, there is link on original post:
http://www.ssdperformanceblog.com/2011/06/fusionio-320gb-mlc-random-write-performance/

Follow my twitter Follow @VadimTk
for further results.

Jun
22
2011
--

Intel 320 SSD random write performance

I have posted results for Intel 320 SSD on our ssdperformanceblog.com blog.
To not duplicate content, there is link on original post:
http://www.ssdperformanceblog.com/2011/06/intel-320-ssd-random-write-performance/

Follow my twitter @VadimTk for further results.

Jun
22
2011
--

Upcoming webinar on Hibernate and Connector/J

On July 12th at 9 AM PST I will be giving a webinar about performance implications for Hibernate and Connector/J. If you cannot attend at this time, a recorded session will be available soon after the webinar.

MySQL is not only about LAMP, and a lot of people use it from Java apps. For some of those, Hibernate is the persistent framework of choice. This webinar will discuss the performance implications of using Hiberate to manage persistence with a MySQL backend, and also some more broad implications for Connector/J  that will apply for anyone using MySQL from Java.

Topics will include:
* Very brief overview of ORMs and Hibernate
* Fetch strategies
* Lazyness
* Manually written SQL
* Concurrency
* Concurrency-related config options for Connector/J

You may register here.

Jun
21
2011
--

Upcoming webinar on Hibernate and Connector/J

On July 12th at 9 AM PST I will be giving a webinar about performance implications for Hibernate and Connector/J. If you cannot attend at this time, a recorded session will be available soon after the webinar.

MySQL is not only about LAMP, and a lot of people use it from Java apps. For some of those, Hibernate is the persistent framework of choice. This webinar will discuss the performance implications of using Hiberate to manage persistence with a MySQL backend, and also some more broad implications for Connector/J  that will apply for anyone using MySQL from Java.

Topics will include:
* Very brief overview of ORMs and Hibernate
* Fetch strategies
* Lazyness
* Manually written SQL
* Concurrency
* Concurrency-related config options for Connector/J

You may register here.

Jun
17
2011
--

Hash Join versus ORM

Today’s installment explains the hash join. It has very different performance characteristics as the nested loops join, requires different indexing, and opens a new challenge for Object-Relational mapping (ORM) tools.

Written by in: Zend Developer |

Powered by WordPress | Theme: Aeros 2.0 by TheBuckmaker.com