Tuesday Nov 04, 2008

Advantages of deploying MySQL database with Solaris Cluster

ritu

The primary advantage of deploying the MySQL database in a Solaris Cluster environment is high availability. The Solaris Cluster environment provides fault monitoring and failover capabilities not only for the MySQL software, but also for the entire infrastructure including servers, storage, interconnects, and the operating system. If any component of the entire infrastructure fails, that failure is isolated and managed independently with no impact on availability.

MySQL Master-Slave configurations, deployed outside of a Solaris Cluster environment, provide limited availability: if the master fails, then the slave can manually be assigned master status and take over operation. However, this process is not automatic but requires manual intervention by a system administrator. Solaris Cluster removes this limitation, as it automatically fails over in the case of a master node failure. In addition, Solaris Cluster provides high availability for slaves as well as for masters. By providing high availability for slaves, these slaves can be kept updated with the masters throughout database transactions, thereby supporting scalability of MySQL database services. The Solaris cluster also provides both failover and scalable Apache Web server instances, thereby offering larger high availability coverage of the SAMP stack. Thus, MySQL Master-Slave configurations deployed in Solaris Cluster environment become highly available in the true sense.

Solaris Cluster deployments provide additional benefits beyond high availability. The Solaris Cluster environment can simplify administration by enabling clustered systems to be managed as if they were a single system. Data services, such as the MySQL database, can be deployed in Solaris Containers, providing the benefits of consolidation (as provided by Solaris Containers) as well as high availability (as provided by Solaris Cluster). Finally, both the MySQL database and Solaris Cluster are free and open source software, helping to contain costs and provide a low-cost solution for highly available databases.

If this is of interest to you , stay tuned here for a pointer on the blueprint for "Deploying MySQL Databases on Solaris Cluster Environment for increased High Availability"

Monday Feb 25, 2008

MySQL Benchmark UltraSPARC T2 beats Xeon on Consolidation of OLTP & Web

Recently we put together a consolidation benchmark to see how an open-source stack performs against the proprietary stack from Microsoft. Solaris, MySQL, and Sun Web Server running on the open-source UltraSPARC T2 processor were pitted against a Microsoft SW stack running on a 4-socket QC Xeon server. This benchmark highlights the continued trend to incorporate MySQL open-source databases and how it works under virtualization (Solaris Zones).

The Sun SPARC Enterprise T5220 (1.4 Ghz UltraSPARC T2 processor) and Solaris Containers managing a consolidation of Open-Source Software components (MySQL Database and Sun Java System Web Server) provided 2.4 times better performance than the HP DL580 system (four Xeon quad-core processors) and a major virtualization software, Microsoft Windows 2003 Server EE, Microsoft SQLserver database and Microsoft IIS webserver.

The Sun SPARC Enterprise T5220 using the MySQL database in Solaris zones is over 3.2 times faster performance than HP DL580 using Microsoft SQLserver database running with the leading virtualization software.

The Sun SPARC Enterprise T5220 using the Sun Java System Web Server with Solaris Zones is 53% faster than than HP DL580 using Microsoft IIS webserver running with the leading virtualization software.

The database performance and the web performance were normalized and equally weighted for an overall performance metric.

The Sun SPARC Enterprise T5220 (1.4 Ghz UltraSPARC T2 processor) including hardware and open-source software costs has 3.7 times better price/performance than the HP DL580 system (four Xeon quad-core processors) with the Microsoft software stack. The Sun solution has 56% less price than the HP/Microsoft/virtualization configuration.

The Sun SPARC Enterprise T5220 (1.4 Ghz UltraSPARC T2 processor) had a 4.1 times better watt/performance than the HP DL580(four Xeon quad-core processors). The Sun system used 73% fewer watts total.

Sun's solution requires half of the rack space of the HP DL580. Sun has an 8.1 SWaP advantage over the HP DL580 (SWaP = Perf /[ Space (RU) x Watts ] )

The Sun SPARC Enterprise T5220 used open-source MySQL 5.0 database software and Sun Java System Web Server 7 which are both free for download to obtain these results.

Performance Comparison:

System Configuration (both 64GB Memory)
CPU RUs OS DB SW Web-
server SW
Sun SPARC Enterprise T5220 1.4GHz 1 chip
8 cores
2 RU Solaris 10 8/07 MySQL 5.0.51a Sun Java System Web
Server 7
HP DL580 G5 2.93GHz 4 QC Xeon
16 cores
4 RU MS Server 2003 EE MS SQL-
server 2005
MS IIS 6.0

System Performance & Cost (server, 64GB, disks, SW)
Web ops/s OLTP txns/m Watts HW+
SW $
Norm
Perf
Watt/
perf
$/perf
Sun SPARC Enterprise T5220 1.4GHz 9546 224K 480W $131k 2.36 203 55
HP DL580 G5 2.93GHz 6250 70K 830W $204k 1.0 830 204
Sun's Advantage 53% 3.2x 73% 56% 2.4x 4.1x 3.7x
Note: Watts are measured server watts during the run.

Benchmark Description

The Full-Stack Consolidated workload consists of 2 primary components: 1) a Web workload consisting of static HTTP requests, and 2) an OLTP database workload using a mix of common SQL transactions executed on a mix of tables in an RDBMS.

The Web workload generates static requests for web pages using a mix of file sizes that range from 100 bytes to 900K bytes with an average file size of 16K bytes. The metrics from this workload includes the number of web operations per second. In this Web workload, a set of 6 client systems emulating thousands of users, generated HTTP requests to all 3 webservers in parallel running on the server.

This performance comparison used an Ad-Hoc OLTP workload, called iGEN OLTP 1.6, which was developed from a realistic customer workload. iGen OLTP avoids problems that plague other OTLP benchmarks like TPC-C. In particular, it is completely random in table row selections and thus is difficult to use artificial optimizations. iGen OLTP stresses process and thread creation, process scheduling, and database commit processing. The database has 10 million customers residing in it, and is approximately 15 GB in size. at time of creation. The transactions are comprised of various SQL transactions: read-only selects, joins, inserts and update operations. For this OLTP workload, 1 client system was used to emulate hundreds of users, generating SQL transactions to all 3 database instances in parallel running on the server.

On the Sun SPARC Enterprise T5220, the 3x Webserver instances and 3x MySQL database instances where isolated on 6 Containers. Each webserver zone shared a single storage array, but separate Gbit network interfaces. The 3x MySQL database zones shared a single network interface but used separate storage arrays for the database files and transaction logs. On the HP DL580, the 3x Webserver instances and 3x SQLserver database instances where isolated on 6 virtual machines. Each webserver VM shared a single storage array, but separate Gbit network interfaces. The 3x SQLserver database VMs shared a single network interface but used separate storage arrays for the database files and transaction logs.

When results listed above were obtained when running both the Web and OLTP workloads concurrently. The performance results for each workload were normalized using the results obtained from the HP DL580 as follows:

Sun SPARC Enterprise T5220 factor = Web + OLTP = 2.36

  1. Web Factor = 9546 ops/sec / (2x 6250 ops/sec) = 0.76
  2. OLTP Factor = 224K txns/min / (2x 70K txns/min) = 1.6

HL DL580 performance factor = Web + OLTP = 1.0

  1. Web Factor = 6250 ops/sec / (2x 6250 ops/sec) = 0.50
  2. OLTP Factor = 70K txns/min / (2x 70K txns/min) = 0.50

The Sun Enterprise T5220 used 3 Solaris Containers to host separate MySQL database instances and 3 Containers to host separate Sun Web Server instances.

Each MySQL and Webserver instance ran in an isolated zone not visible o any processes or users in the other containers.

The combination of OLTP and Web workloads consumed an average of 95% of cpu utilization when running concurrently.

System Configuration & Results Summary

Sun SPARC Enterprise T5220 (1 chip, 8 cores)9546 Web operations/sec, 224K OLTP Txns/min.
HP DL580 (4 chips, 16 cores)6250 Web operations/sec, 70K OLTP Txns/min. Results as of 2/22/2008.

Results
Sun SPARC Enterprise T5220 9546 Web ops/sec, 224K OLTP Txns/min
HP DL580 G5 6250 Web ops/sec, 70K OLTP Txns/min
Reference Date: Feb 22, 2008
Systems: Sun SPARC Enterprise T5220
HP DL580 G5
Processor/GHz of Server: 1x UltraSPARC T2 1.4 GHz,
4x Intel Xeon X7350 2.93 GHz
Operating System/Virtualization: Solaris 10 8/07 & Solaris Containers
Microsoft Windows Server2003 EE & a leading virtualization platform

Sun SPARC Enterprise T5220 with:

  • 1x UltraSPARC T2, 1.4  GHz 8-core processor
  • 64 GB of memory
  • 4x Internal 146GB SAS disks
  • 2x 750W Power Supplies
  • 4x Sun StorageTek ST2540 arrays (12x 146GB disks per array)
  • 2x Sun StorageTek 4Gb Dual-Port Fibre Channel PCI-E HBA (from Emulex)
  • 1x Sun PCI-E Dual Giga Ethernet Adapter UTP
  • Solaris 10 8/07
  • Solaris Containers
  • MySQL 5.0.51a
  • Sun Java System Web Server 7 u1

HP DL580 G5 with:

  • 4x Intel Xeon X7350, 2.93  GHz 4-core processors
  • 64 GB of memory
  • 5x Internal 146GB SAS disks & Smart Array P400i SAS Ctlr
  • 4x 800/1200W Power Supplies
  • 4x Sun StorageTek ST2540 arrays (12x 146GB disks per array)
  • 2x QLogic SANblade QLE2462 PCI-E Dual Port FC adapter
  • 3x HP NC360T PCI Express Dual Port Gigabit Server Adapter
  • Leading Virtualization Platform
  • Microsoft Windows Server 2003 EE 64-bit
  • Microsoft SQLserver 2005
  • Microsoft IIS 6.0 Webserver
About

Ritu Kamboj

Search

Categories
Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today