Keeping SharePoint Always On
- 1. Antonio Maio Protiviti, Senior SharePoint Architect & Senior Manager Microsoft SharePoint Server MVP Keeping SharePoint Always On Email: Antonio.maio@protiviti.com Blog: www.trustsharepoint.com Slide share: http://www.slideshare.net/AntonioMaio2 Twitter: @AntonioMaio2
- 2. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. About Protiviti INDIA (3) Protiviti (www.protiviti.com) is a global consulting firm that helps companies solve problems in finance, technology, operations, governance, risk and internal audit. Through our network of more than 70 offices in over 20 countries, we have served more than 35 percent of FORTUNE® 1000 and Global 500 companies. We also work with smaller, growing companies, including those looking to go public, as well as with government agencies. Protiviti is a wholly owned subsidiary of Robert Half International Inc. (NYSE: RHI). Founded in 1948, Robert Half International is a member of the S&P 500 index. • 3,200+ professionals • 1,000+ clients • 70+ offices • Over 20 countries in the Americas, Europe and Asia-Pacific Protiviti is one of the fastest growing consulting firms worldwide. Our revenues have increased from US $15 million in 2002, to US $423.8 million in 2011.
- 3. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Goal Educate on Building a SharePoint Environment to Optimize Performance and Resilience
- 4. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Agenda • Baseline – Front End, Back End, Network, Bandwidth, User Access – Time to 1st Byte • Infrastructure Considerations – System Architecture – Topology – SQL Server • High Availability • Disaster Recovery
- 5. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Baseline Architecture: Front End • Memory: 50-60% Utilization (Memory/Available Bytes Counter) • CPU Usage: 30-50% Average Usage • Disk I/O: Disk Queue Length (Differs Per Environment) • Concurrent Connections: Depends on user base, 150-250 per server – Larger total number of users – Dependent on what users are doing (public site vs intranet very different)
- 6. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Baseline Architecture: Back End • Memory: 30-60% Utilization (Memory/Available Bytes Counter) • CPU Usage: 30-50% Average Usage • Disk I/O: Disk Queue Length = Physical Disk(_Total)/Current Disk Queue Length Counter • Concurrent Connections / SQL Blocking: SQL Server: Locks (_Total)Number of Deadlocks/sec counter – Acceptable: < 1 per second • Fragmentation: < 8% per server • Traffic: Realistic Number of Page Hits – MS IT caters to 6 million hits per day
- 7. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Baseline Architecture: Data Center • Cross Data Center Farm • High Bandwidth • Fiber Optic Connection – >= 1 gigabit per second bandwidth – <1 millisecond one-way ping response • Load Balanced End User Requests • Single Data Center Farm • High Network Speed: Gigabit Speeds Ideal • <= 1 millisecond response between all servers • Load Balance End User Requests
- 8. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Baseline Architecture: Network • Use Case: • Server: Receive and Process initial Request • Server: Request and Transfer • Client: Response rendered • Bandwidth: Amount of data that can be sent through the wire • Latency: Time taken to travel from source to destination
- 9. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. #Users # Concurrent Users Local Network Access WAN/Remote Access 100 – 5000 10 – 500 Bandwidth: = 3Mbps Latency: < 100 ms Bandwidth: = 1.5Mbps Latency: < 100 ms 10,000 1,000 Bandwidth: >3Mbps Latency: <250 ms Bandwidth: >1.5Mbps Latency: <250 ms 100,000 10,000 Bandwidth: > 3Mbps Latency: < 250 ms Bandwidth: > 1.5Mbps Latency: < 250 ms Baseline Architecture: User Access • Bandwidth Table 1.5 MB data per second < 100 ms latency
- 10. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Baseline Architecture: Network User Access • Test Ping Access - Ping each of the server in the “server list” to ensure that all servers are reachable - Connect to the SQL server to verify DB connectivity - Run a 10 minute ping test against each to ensure that the environment meets the <1 ms latency requirement - Generate output to show the percentage of pings that complete in 5 ms, 4 ms, 3 ms, 2 ms and 1 ms respectively • Test Network Latency – http://blogs.msdn.com/b/erica/archive/2013/11/11/sharepoint-2013-network-latency-test-script.aspx
- 11. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Baseline Architecture: Time to 1st Byte • See TechNet: http://technet.microsoft.com/en- us/library/cc964305.aspx • $siteUrls = @( “http://sharepoint.com”, “http://my.sharepoint.com”) • $siteUrls | Ping-Url
- 12. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Server Specifications Size Matters!
- 13. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Server Specifications Small Farm • RAM: => 8 GB • Processor: 64-bit, => 4 cores • Hard Disk Space • 80 GB (system drive) • Other disk space as needed Medium Farm • RAM: => 16 GB • Processor: 64-bit, => 8 cores • Hard Disk Space • 80 GB (system drive) • Other disk space as needed Large Farm • RAM: • Up to 2 TB Data = 32 GB • Between 2 TB and 5 TB = 64 GB • > 5 TB = 64 GB or higher • Processor: 64-bit, => 8 cores • Hard Disk Space • 80 GB (system drive) • Other disk space as needed
- 14. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: SQL Server • Cluster – Expensive, Complex (Single Node), Shared Storage, Fiber / SCSI • Mirroring – Isolated Storage, Single Mirrors, Not Grouped and needs Snapshots • Log Shipping – Complex, Database unusable until recovery state and logs applied / updated • Always On Availability Groups – Multiple Copies of Databases, Isolated Storage, Off Server Backup, Readable Secondary's and Grouped Databases
- 15. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Topology
- 16. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Topology One Server Farm Two Tier farm
- 17. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Topology Four Server Farm
- 18. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Topology Six Server Farm
- 19. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Topology Six Server Farm Load Balancer
- 20. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Topology Large Farm Web servers for all incoming requests Dedicated web server(s) for crawling and administration Query and index servers All other search components Servers for running sandboxed codeAll other services (use these servers for the Central Admin site) Content databases and configuration database All other SharePoint databasesSearch databases
- 21. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Topology Hybrid Farm Load Balancer
- 22. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Infrastructure: Topology Always On – Availability Group On-PremiseCOLO 10.10.1.106/24 10.10.1.107/24 192.168.1.103/24 192.168.1.102/24 192.168.1.101/24 192.168.1.100/24 10.10.1.103/24 10.10.1.104/24 10.10.1.101/24 10.10.1.102/24 10.10.1.100/24 10.10.1.254/24 192.168.1.254/24 192.168.1.120/24 10.10.1.120/24 VIP Address
- 23. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. High Availability & Disaster Recovery
- 24. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. High Availability & Disaster Recovery • Goal of a high availability - to minimize or mitigate downtime • High Availability measured by end user’s experience and expectations • Business impact of downtime – Measured in the cost of information loss, property damage, decreased productivity, opportunity costs, or contractual damages – Availability = Actual uptime/Expected uptime X 100%
- 25. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. A Disaster Recovery strategy must be based on metrics, defined by determining the risks and downtime cost to the business if a disaster occurs, metrics such as: Recovery Time Objective (RTO) The duration of an outage. Initial goal is to get the system back online in (at least) a read-only state to allow for investigation of the failure. The ultimate goal is to restore full service of the system or application. 1 Recovery Point Objective (RPO) A measurement of the amount of data which the organization is prepared to lose in the event of an outage. 2 3 Recovery Level Objective (RLO) The granularity with which data must be recoverable. If the recovery objective requires the entire farm, web application, site collection, site, list/library, item or version. Disaster Recovery
- 26. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Determining downtime costs is a key to identifying the metrics needed to meet the business goals related to disaster recovery, and the strategy and systems required to achieve those metrics. Downtime costs can be calculated by identifying the major aspects of an outage. • Root Cause Analysis of Downtime • Types of Loss Expected by the Business • Impacts and Related Business Factors • Direct and Indirect Causes • Appropriate Response Activities Determining Downtime Costs
- 27. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Selecting a standby data center is a critical decision in developing a Disaster Recovery strategy. Its required when local redundant systems and backups cannot be used to recover an outage at a primary data center. Cold Standby A secondary data center providing availability within hours or days. Warm Standby A secondary data center providing availability within minutes or hours. Hot Standby A secondary data center providing availability within seconds or minutes. Selecting a Standby Data Center Best Practices
- 28. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Baseline Architecture: Thoughts • Front End – Run IIS version 7.0+ on 64-bit servers – Use a front-end and back-end NIC configuration for IIS – Load balance client traffic – Use IIS compression for static content – Enable caching (BLOB cache) • Back End – Limit database size to enhance manageability – Allocate storage for versioning and the recycle bin – Manage large lists for performance – Separate and prioritize data among disks and create disk groups for specific data
- 29. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. Final Thoughts • Establish a baseline for performance comparison • Make data-driven decisions • Simplify topologies • Periodically standardize environment configuration • Plan strategy for High Availability and Disaster Recovery • Routinely verify and audit system configuration – Use different teams or individuals for auditing • Monitor all aspects of the environment – System Center / http://gallery.technet.microsoft.com/scriptcenter/2d537e5c- b5d4-42ca-a23e-2cbce636f58d/
- 30. © 2014 Protiviti Consulting Private Ltd. An Equal Opportunity Employer. • 7th Annual Conference • 6 SharePoint MVPs presenting • Full-Day Training Workshops pre and post conference • Separate tracks for Business Users, Technical Users & UX Professionals • Government, Association, Nonprofit & Group Discounts Available Get 15% off! Use: PRO15 when registering Learn More & Register at SharePointConference.org Dallas, TX October 20, 2014
- 31. Antonio Maio Protiviti, Senior SharePoint Architect & Senior Manager Microsoft SharePoint Server MVP Thank You! Email: Antonio.maio@protiviti.com Blog: www.trustsharepoint.com Slide share: http://www.slideshare.net/AntonioMaio2 Twitter: @AntonioMaio2