3-Steps to Deploying Mission Critical Web & Telecoms Services

There are multiple architectures that can be used to achieve highly available database services, each differentiated by the levels of uptime they offer.  These architectures can be grouped into three main categories:
- Data Replication
- Clustered & Virtualized Systems
- Shared-Nothing, Geographically-Replicated Clusters

As illustrated in the figure below, each of these architectures offers progressively higher levels of uptime, which must be balanced against potentially greater levels of cost and complexity each incurs. 

Simply deploying a high availability architecture is not a guarantee of actually delivering HA.  In fact, a poorly implemented and maintained shared-nothing cluster could easily deliver lower levels of availability than a simple data replication solution.

 
9s.png



To reduce risk and accelerate the evaluation process, we've recently developed some new resources that demonstrate how to get started with the MySQL Cluster database - a key option for users needing to deliver MySQL-backed solutions with 99.999% availability (i.e. less than 5 minutes downtime per year).

In this 3 x step series, we've shared best practices in installing / configuring / testing MySQL Cluster, then moving from evaluation into production deployment and practical guidance in optimizing application performance. 


•    Part 1: MySQL Cluster Deployment Best Practices
Listen to the On-Demand Webinar

•    Part 2: Getting Started with MySQL Cluster
Listen to the On-Demand webinar

•    Part 3: Guide to Optimizing the Performance of the MySQL Cluster Database
Download the Whitepaper

And of course you can feedback your experiences on the MySQL Cluster mailing lists and user forum

Happy clustering !

Comments:

Hi Mat

Nice update to old graph. However, note that "shared nothing" doesn't add anything to availability, it helps with scalability.

Ok, so if you have now added the possibility to let a partial cluster continue running even if some partitions are missing, then I guess it adds to availability too. Even so, it still feels weird.

I guess the fact that MySQL Cluster's failover time is an order of magnitude faster than RAC is an aspect of availability too, but I wouldn't know if that is a property of the shared nothing architecture, or just better engineering in general.

Posted by Henrik Ingo on September 30, 2010 at 01:16 PM GMT #

Hey Henrik
re the "old graph", that is the nicest accusation of plagarism I've ever heard!...and partially true !

The ability to withstand network partitions was one aspect to shared-nothing availability, but it is not alone.

Also:
- no shared storage or locks to worry about, therefore failover times are much faster. In a good configuration, typically less than a second, versus 60 seconds or more on many shared disk systems. A few of those per year, and you are already outside of 5 x 9s on failover time alone

- rolling upgrades covering systems and database maintenance, eliminating planned downtime that would impact availability on shared-all systems

Posted by Mat Keep on September 30, 2010 at 11:34 PM GMT #

Post a Comment:
  • HTML Syntax: NOT allowed
About

Get the latest updates on products, technology, news, events, webcasts, customers and more.

Twitter


Facebook

Search

Archives
« April 2014
SunMonTueWedThuFriSat
  
2
5
6
9
10
11
12
13
15
16
17
18
19
20
23
24
25
26
27
28
29
30
   
       
Today