Distributing traffic smartly in a small cluster setup can be enough to deliver content fast and easily. In this session I would like to show our setup on tradebit.com, where we host thousands of merchants and over 60 Terabytes of downloadable products on a very simple and easy self-made LAMP cluster with SphinxSearch Server.
The session includes a complete sketch of all involved servers, NFS response times, traffic handling methods with Squid reverse proxies, catalog browsing with MySQL and Sphinx and how the data is distributed with 2 masters and 6 slaves with MySQL 5.x.
In his Keynote Peter Zaitsev looks at MySQL Ecosystem state in 2012. In this keynote you will learn about the market trends affecting MySQL ecosystem, new products and industry practices you need to know about to achive maximum success with your MySQL installation.
MySQL's configuration file is often the focus of too much attention and too much tweaking of variables that make no difference -- or worse, have the potential to negatively impact performance. The sample default configuration files that come with MySQL are unfortunately not very helpful or good, either. You'll learn a practical approach to generate a sensible configuration file that sets what's needed and omits what isn't. The result will be a more stable and performant server that's resistant to many common problems.
Percona Toolkit is a collection of command-line tools designed to help MySQL DBAs perform diagnostic and maintenance tasks more efficiently. This session will begin by discussing some of the problems that are harder to solve in MySQL than in other RDBMS. We'll then review each problem case and present a Percona Toolkit tool that can help solve it, concluding with a review of upcoming changes in MySQL development that may simplify the resolution of some of the described problem cases.
This talk is about the new High Availability solution in MySQL arena: Percona XtraDB Cluster.
Percona XtraDB Cluster provides:
1. synchronous replication
2. supports multi-master replication
3. parallel applying AKA “parallel replication”,
4. automatic node provisioning.
5. primary focus on data consistency
I will cover how to install it, add nodes to cluster and make it running.
This session talks about a case study which focuses on some recent challenges which we faced with our internal Web portal built on Drupal & mysql and how the transformation in thought process of the individuals brought a change in technology resulting in some tremendous performance gains of upto 70% and laying a solid foundation to build a robust website on. The session will explore the technical operational framework and partnership across multiple teams which led to the success of the website by achieving sub second response times.
Stored Routines (procedures and functions) were first added to MySQL 5.0. Stored routines in other RDBMS systems have been blamed for poor performance, in addition to being hard to manage and maintain. This talk will focus on stored routines in MySQL 5.0 and share some performance data and best practices for using stored routines smartly and effectively. Come learn why, how, when, and where to best use stored routines, and when to avoid them.
As organizations deploy more of their infrastructure into Amazon's Web Services, there is a growing need for guidance on scaling MySQL in the cloud. This presentation will address performance challenges and management issues in the cloud. By the end of the presentation, attendees will understand cloud-specific advantages, scaling challenges and how to achieve high availability on that environment.
As your project and codebase grows and matures, you often find that code you had written before just isn't supporting what you hope to accomplish now. At first it may be restricted to a few if...else blocks, but soon this grows to be entirely unmanageable and makes it difficult to adapt to support more complex database backends. With a full re-write difficult and costly to do in many cases, the next best approach is a step by step refactor.
It is a myth that website that hosts more than the tiniest amount of visitors must run on dedicated hardware. More often than not, servers are completely misconfigured to make the optimal use of their available hardware resources. When your web server, your scripting language and your database are all on the same server, and that server has limited resources, it's even more important to tune things correctly.
MySQL-based systems are growing into mixed environments that include MySQL along with other SQL and NoSQL databases connected by complex replication topologies that span multiple sites. Administering such environments has become a time-consuming and almost impossible task. This talk covers how Continuent Tungsten handles these problems and reduces the stress on overloaded DBAs.
When using mysql for large data (more than time times bigger than main memory) these these challenges often arise: loading data fast; maintaining indexes under insertions deletions, and updates; adding and removing columns online; adding indexes online; preventing slave lag; compressing the data effectively. I'll show why some of these challenges are difficult to solve using storage engines based on B-trees, and how Fractal Tree data structures work and why they can help solve these problems.
Galera is groundbreaking new replication technology powering clustering solutions like Percona XtraDB Cluster. Galera gives you synchronous replication, which is a powerful tool changing the rules in the game when designing HA application stacks. With synchronous replication all database nodes can be treated equally and this simplifies the HA architecture remarkably - the complexity is boxed in cluster implementation.
But, how to make best use of synchronous replication? This presentation will answer questions,like: How to organize the synchronous cluster topology?...
Once MySQL users discover dual-master replication, they are often tempted to allow their applications to write to both at the same time. This talk will explore the dangers of doing so as well as the (limited) possibilities it allows.
Illustrated examples will be used to show how this breaks down ACID compliance at fundamental levels.
This talk would be targeted to beginner/intermediate MySQL Users and DBAs.
This talk will describe the processes we used at New Relic to upgrade from 5.0 to 5.5. Since we provide real time application performance analytics we could not afford any lengthy downtime. We also had some interesting technical restraints that had to be overcome.This talk will describe the planning, testing, replication upgrade processes.
Vitess is an open source project that packages many of the ad-hoc processes and conventions that grew out of managing and scaling MySQL at YouTube. Vitess is now at the core of our MySQL serving infrastructure. In this session, we'll cover our vision of where the project is headed as well as what we've achieved so far.
Schema changes on live systems are hard. They are inherently disruptive in Mysql. We at Etsy have devised a way to make schema changes in a non-disruptive way, and we routinely perform them without any user-facing impact. I will give a brief description of our Mysql architecture, followed by a fairly detailed walk-through of our schema changes procedures. Finally, I will present the tooling we've built around our schema changes procedures: our Schemanator tool.