Jak na… – plánování hardwaru pro OpsMgr 2007 (OpsMgr 2007 Hardware Guidance)

Již na začátku dubna uveřejnil Satya Vel na blogu produktového týmu Operations Manager doporučení pro nasazení OpsMgr 2007. Při čekání na finální verzi nástroje System Center Capacity Planner, určeného pro plánování a konfiguraci hardware pro nasazení OpsMgr 2007, sepsal níže popsané body.

Dnes je sice SCCP již k dispozici, ale v komplexnějším prostředí pro implementaci je lepší více plánovat než následně složitě řešit výkonostní problémy. Takže zde to je (převzato z blogu SV).

….

Some additional deployment tips:

  • RAID Controller needs to have Write Cache enabled with a ratio of Read/Write set to 50/50 or 25/75
  • We recommend a maximum of 2000 Agents in a single Management Server (MS)
  • We recommend a maximum of 800 Agents in a single Gateway Server (GW)
  • We do not recommend agents reporting to MS when there are GW Servers reporting to MS
  • We do not recommend agents reporting to RMS when there are MS Servers reporting to RMS
  • If you are having multiple Management Groups write into a single Data Warehouse then I would add the recommended RAM requirements for each configuration
  • Always use 64bit servers for the Root Management Servers
  • Having all the OpsMgr roles on a single server is recommended if you are either testing the product or building a pre-production environment.
  • Do not have agents report directly to the Root Management Server. The RMS computes and distributes configuration information to all of the Management Servers and agents in the Management Group and therefore it is important not to add additional load on the RMS.
  • The number of simultaneous consoles has a direct impact on the performance of the system. Having a high end server for the Root Management Server will provide greater console scale and fast console response time.
  • The disk volume for the database should typically be RAID 10 with an appropriate number of spindles.
  • On the Root Management Server the most critical resource is RAM followed by CPU.
  • Use 64-bit hardware and a 64-bit operating system to allow for easy increases in memory beyond 4 GB and provide room for growth as needed.
  • For any given deployment it is a best practice to have the Root Management Server and Management Server in close network proximity to the database and data warehouse as they both write directly to each one.
  • Consider placing the SQL data file and transaction log on separate arrays. This placement allows the transaction log volume to perform I/O more efficiently because its workload consists of mostly sequential writes.
  • Use a battery-backed write-caching disk controller. Workload on the Reporting data warehouse and Root Management Server is reduced by write-caching on disk controllers.
  • When configuring read caching vs. write caching on disk controllers for the Root Management Server, allocating at least 50% of the cache to write caching is recommended.
  • It is recommended that the SQL Server tempdb database file and transaction log file be re-sized to 20% of the total size of the Operations Manager Database and Operations Manager Data Warehouse databases. It is also recommended to put the tempdb on a high-performance subsystem.

– Satya Vel

Reklamy

Zanechat Odpověď

Vyplňte detaily níže nebo klikněte na ikonu pro přihlášení:

WordPress.com Logo

Komentujete pomocí vašeho WordPress.com účtu. Odhlásit / Změnit )

Twitter picture

Komentujete pomocí vašeho Twitter účtu. Odhlásit / Změnit )

Facebook photo

Komentujete pomocí vašeho Facebook účtu. Odhlásit / Změnit )

Google+ photo

Komentujete pomocí vašeho Google+ účtu. Odhlásit / Změnit )

Připojování k %s

%d bloggers like this: