Designing a Microsoft SharePoint 2010 Infrastructure Vol 1 part 15 potx

MCT USE ONLY. STUDENT USE PROHIBITED Planning for Performance and Capacity 3-7 Performance of Web Front End Servers Key Points WFE servers form the SharePoint connection point for clients that request content or services. This means that all client requests place some load on the WFE servers. WFE servers render pages before returning requested pages to a browser. In small farms, WFE servers often perform application roles in addition to the WFE role. WFE servers do not require large quantities of disk storage, but rely heavily on processor and memory for performance. The following table describes the processor and memory load characteristics for WFE servers. Service application or feature CPU load Memory load SharePoint Foundation Service High High Timer Service Medium Medium Logging Service Medium Medium MCT USE ONLY. STUDENT USE PROHIBITED 3-8 Designing a Microsoft® SharePoint® 2010 Infrastructure Service application or feature CPU load Memory load User Profile Service Low Medium Word Viewing Service Low Low PowerPoint Service Medium Medium Excel Calculation Service Medium Low Visio Service Low Low Access Service Low Low Managed Metadata Service Low Medium Web Analytics Service Low Low Business Connection Service Medium Medium InfoPath Forms Service Medium Medium Word Conversion Service Low Low PerformancePoint Service Medium Medium Sandboxed Solutions Low Low Workflow Capabilities High High Note: These load characteristics exist even when all service applications are running on dedicated application servers. For example, the Excel Calculation Service will increase CPU usage on the WFE server, even if you have an application server dedicated to running Excel Services. You can use the table as a guideline to help when you calculate workload for WFE servers. Use the following information when calculating server workload: • Consider that medium-cost CPU operations are three times heavier on workload than low-cost CPU operations. • Consider that high-cost CPU operations are five times heavier on workload than low-cost CPU operations. MCT USE ONLY. STUDENT USE PROHIBITED Planning for Performance and Capacity 3-9 To improve performance of page rendering and client access, you can add more WFE servers to the farm and implement network load balancing. Additional Reading For more information about sizing for SharePoint Server 2010, see http://go.microsoft.com/fwlink/?LinkID=200850&clcid=0x409. MCT USE ONLY. STUDENT USE PROHIBITED 3-10 Designing a Microsoft® SharePoint® 2010 Infrastructure Performance of Application Servers Key Points Application servers host service applications. Service applications offer a more flexible approach to farm services than previous versions of Microsoft Office SharePoint; you now have more options for controlling which servers in the farm run specific service applications. Different service applications have different workload profiles, but you can dedicate specific servers to specific service applications. You can also scale out by specifying multiple servers for a specific service application. Different service applications have different workload profiles, but most application services do not require local storage on the application server. The main hardware requirements for application servers are processor and memory. The following table describes the processor and memory load characteristics for application servers. MCT USE ONLY. STUDENT USE PROHIBITED Planning for Performance and Capacity 3-11 Service application or feature CPU load Memory load SharePoint Foundation Service None None Timer Service Medium Medium Logging Service None None User Profile Service Medium Medium Word Viewing Service High Medium PowerPoint Service High Medium Excel Calculation Service Medium High Visio Service High High Access Service High Medium Managed Metadata Service Medium Medium Web Analytics Service None None Business Connection Service High High InfoPath Forms Service Medium Medium Word Conversion Service High Medium PerformancePoint Service High High Sandboxed Solutions High High Workflow Capabilities None None Generally, for best performance, server roles should not share service applications with a high load. Consider that some service applications will also increase the workload on the WFE servers. Additional Reading For more information about performance test results and recommendations, see http://go.microsoft.com/fwlink/?LinkID=201227&clcid=0x409. MCT USE ONLY. STUDENT USE PROHIBITED 3-12 Designing a Microsoft® SharePoint® 2010 Infrastructure Performance of Search Servers Key Points Search is typically a significant benefit of implementing SharePoint 2010, but it places a large workload burden on the farm. When you consider farm performance, you must often consider search performance specifically in the context of the farm. Search servers can perform two functions: • The crawl component crawls and indexes content primarily in the SharePoint content databases, although it can also index other types of storage repository. The crawl role builds the index and submits index updates to the search query role. Crawl components aggressively use CPU bandwidth. Optimally, a given crawl component can utilize four CPU cores. Memory is not as critical for the crawl component. • The query component responds to user search requests. When users enter a search in a SharePoint site, SharePoint 2010 submits the query to a server that hosts the query role to return a result set. All servers that host the query role have a copy of the index that the crawl role generates. However, additional MCT USE ONLY. STUDENT USE PROHIBITED Planning for Performance and Capacity 3-13 search databases also hold specific search information. The query role typically has high processor and memory requirements. Note: A single search server can perform both crawl and query functions. In addition, the Search service has a significant impact on database servers. The Search service places a high load on CPU, input/output (I/O), and storage components. Crawl databases aggressively use I/O bandwidth. A crawl database needs 3,500 I/O operations per second (IOps) for crawling activities; it will consume as much as 6,000 IOPS, based on the available bandwidth. Question: Can you spread the crawl component across multiple servers in a farm? MCT USE ONLY. STUDENT USE PROHIBITED 3-14 Designing a Microsoft® SharePoint® 2010 Infrastructure Performance of Database Servers Key Points Database servers store most SharePoint content and most SharePoint configuration data. Not all service applications affect database servers, because some service applications do not require databases. However, storage access times and storage capacity are a key focus for the role. The following table describes the processor, I/O load, and storage characteristics for database servers. Service application CPU load I/O load Storage SharePoint Foundation Service Medium High High Timer Service None None None Logging Service Medium High High User Profile Service High High Medium Word Viewing Service None None None MCT USE ONLY. STUDENT USE PROHIBITED Planning for Performance and Capacity 3-15 Service application CPU load I/O load Storage PowerPoint Service None None None Excel Calculation Service None None None Visio Service Low Low Low Access Service Low Low Low Managed Metadata Service Low Low Medium Web Analytics Service High High High Business Connection Service None None None InfoPath Forms Service Low Low Low Word Conversion Service Low Low Low PerformancePoint Service Low Low Low Sandboxed Solutions None None None Workflow Capabilities None None None Disk Storage Disk storage types and redundant array of independent disks (RAID) configuration can have a significant impact on the performance of database servers. Direct attached storage (DAS) is a configuration where a storage controller without a network interface connects the server and disk hardware. DAS typically uses Small Computer System Interface (SCSI), Serial Attached SCSI (SAS), or Serial ATA (SATA) disks. A storage area network (SAN) connects server hardware to disk hardware over a dedicated network. This can be an Ethernet-based network or a Fibre Channel– based network. SAN storage appears to the operating system as locally attached disks. SANs offer high-speed, high-capacity storage with additional capabilities, such as the ability to support SQL Server clusters. However, they create a significant cost increase over DAS. Network attached storage (NAS) is a self-contained storage device that a server or client can access over the network, typically through traditional file-share methods. MCT USE ONLY. STUDENT USE PROHIBITED 3-16 Designing a Microsoft® SharePoint® 2010 Infrastructure Note: Only content databases that use remote binary large object (BLOB) storage support NAS storage. Any network storage architecture must return the first byte of data within 20 ms. You can use different disk types and different RAID configurations to support specific performance requirements. SAS disks typically support faster access times, although specific RAID configurations—such as RAID 1, RAID 5, or RAID 10—can also affect read or write access times. Note: You should consult the guidelines that the storage hardware manufacturer provides to determine the RAID configuration effect on performance. If you experience performance or storage bottlenecks with the database server role, you can add more database servers to the farm and spread your database requirements across multiple servers. Question: Why should you consider RAID storage options for your database and transaction log file storage? . server and disk hardware. DAS typically uses Small Computer System Interface (SCSI), Serial Attached SCSI (SAS), or Serial ATA (SATA) disks. A storage area network (SAN) connects server hardware. most SharePoint content and most SharePoint configuration data. Not all service applications affect database servers, because some service applications do not require databases. However, storage. has a significant impact on database servers. The Search service places a high load on CPU, input/output (I/O), and storage components. Crawl databases aggressively use I/O bandwidth. A crawl