Setting up a Data Center Guidelines


Tier 1
Single path for uplink, servers, power & cooling distribution w/o redundant components & may not have an engine generator, guaranteeing 99.671% availability (28.8 hours of downtime).

NOTE: The data center must be shut down for annual predictive maintenance & repair work. Operation errors or spontaneous failures of infrastructure components will cause a data center disruption. An example of a Tier 1 data center may be a small businesses where IT is intended for internal business processes.

Tier 2
Tier 1 + Single path for cooling distribution w/redundant components & has UPS & engine generators but their capacity design is N+1, w/o single power path, guaranteeing 99.741% availability (22 hours of downtime).

NOTE: Maintenance of the critical power path & other parts of the site infrastructure will require a shutdown of computer processes. An example of a Tier 2 data center may be internet-based companies w/o serious financial penalties for QoS commitments.

Tier 3
Tier 1 + Tier 2 + Multiple active power & cooling distribution paths, but only one path active, which has redundant components, and is concurrently maintainable. Dual-powered equipment & multiple uplinks, guaranteeing 99.982% availability (1.6 hours of downtime).

NOTE: This topology allows for any planned site infrastructure maintenance activity w/o disrupting the computer systems operation in any way. An example of a Tier 3 data center would include companies that span multiple time zones or whose IT resources support automated business processes.

Tier 4
Tier 1 + Tier 2 + Tier 3 + Multiple simultaneous active power & cooling distribution paths, has redundant components, and is fault tolerant. All components are fully fault-tolerant including uplinks, storage, chillers, HVAC systems, servers, etc. This fault-tolerant architecture provides the ability of the site infrastructure to sustain at least one worst-case unplanned failure or event with impact to the critical load. This typically includes a System+System topology. Everything is dual-powered, guaranteeing 99.995-99.999% availability (at most 0.4 hours of downtime).
NOTE: Typically, the better Tier 4 Data Centers include the following:
Location

  • Strategically located on Airport power grids to prevent rolling blackouts as airport grids are the last to go and are more protected.
  • They have 3-day fuel reserves on-site and same-day contracts with fuel providers for helicopter transports.

Data Replication

  • In addition to all the redundancy & backups within the data center there are real-time site-to-site replications generally in different geographical regions & climates. This way if one data center goes down completely the end-users never know.

Multi-tenant platform security

  • Multiple redundant, enterprise-class firewall systems integrate firewall, VPN and traffic management to prevent unwarranted intrusions and ensure only authorized users access the environment. IDS (Intrusion Detection System) to detect malicious network traffic and computer usage that often cannot be caught by a conventional firewall.

Physical Security

  • Dedicated, full-time security staff monitoring & guarding the data center 24x7x365 w/sophisticated pan/tilt closed-circuit TVs, led by a CISSP (Certified Information Systems Security Professional) analyst. Secure access is strictly enforced using the very latest technology, including electronic man-trap devices between lobby & datacenter, bio-metric authentication, motion sensors, & controlled ID key-cards. Security guards are stationed at every entrance to the site.
  • Every employee, regardless of their role, undergoes a rigorous background check.
  • Employee access to passwords, encryption keys and electronic credentials is also strictly controlled.

Compliance

  • SAS70 Type II certification. It is a rigorous certification requiring an independent auditor to validate, in their opinion, that the controls and processes in use are effective over its one year evaluation period.
  • SOC 2 & 3 are the new version reports of SAS70 Type II, which are being phased out as of 6/15/11. They more directly address all of the 5 concerns: Security, Availability, Processing Integrity, Confidentiality, & Privacy.
  • PCI DSS. This assures credit card payments processed through a hosted site are secure from fraud and data compromise.

NOTE: Examples of a Tier 4 requirement include companies who have extremely high-availability requirements for ongoing business such as E-commerce, market transactions, or financial settlement processes.

Aspects to consider in Data Center Site Selection:
Location

  • Earthquake Zone
  • Flood Plains
  • Hurricanes or Tornadoes
  • Proximity to Major Highways
  • Proximity to Railway Lines
  • Proximity to Hazardous Areas
  • Proximity to Airports or Flight Corridors

Infrastructure

  • Availability of Electrical Capacity
  • Availability of Diverse Power Feeders
  • Utilities Expansion/Upgrades
  • History of Outages

Water

  • Diverse Source Supplies
  • Water Storage

Communications

  • Availability of Diverse Carriers
  • Availability of Diverse Services
  • Physical Security
  • Alarms & Monitoring

Economics

  • Land
  • Construction
  • Utilities
  • Labour
  • Communications

Staffing

  • Accessibility
  • Public Transportation
  • Recreational Facilities
  • Housing
  • Amenities

Aspects to consider in Data Center Site Evaluation:

  • Electrical
  • Utility Service
  • Lightning Protection
  • Power Backbone
  • UPS Systems
  • UPS Batteries
  • Engine Generator
  • Load Bank
  • Critical Power Distribution
  • Grounding

Mechanical

  • Raised Floor Cooling
  • UPS Cooling
  • Mechanical Plant

Support Systems

  • Contamination
  • Fire Detection & Protection
  • Physical Security
  • Alarms & Monitoring

Standards & Compliance

  • SOC 2 or SAS70, Type 2
  • TIA-942
  • PCI DSS

There are a few different ways to rank the aspects that comprise a data center. Syska Hennessy Group offers a method, unlike the four-tiered system from the Uptime Institute, which examines 11 different aspects of performance & measures them on a scale from 1 to 10.
Below are the 11 items that Syska Hennessy Group ranks:
1.      Power
2.      HVAC
3.      Fire & Life Safety
4.      Security
5.      IT Infrastructure
6.      Controls & Monitoring
7.      Commissioning & Testing
8.      Operations
9.      Maintenance
10.      Operations & Maintenance Procedures, And
11.      Disaster Preparedness.
Here are some good references: http://uptimeinstitute.com and http://www.syska.com
In the meantime while you are figuring which route you want to go down, you might want to look into SonicWALL’s CDP (Continuous Data Protection) offering. They have two flavors: a) you can host the data yourself or b) you can use their data center and have them carry the liability. The great thing about their solution is that it’s a policy-based backup, Site-to-site LAN-to-WAN or LAN-to-LAN w/Universal system recovery (BMR) and offsite/DR backup all-in-one with central management and remote management should a local site location go down. It has built-in AES 256-bit encryption in data transit & at rest. It uses fileset backup methodology combined w/data de-duplication & advanced granular versioning & trimming to efficiently capture catalogues and preserves each backup in chronological versions while excluding duplicate data. Bandwidth throttling…point/click recovery…the list goes on. We have had a lot of success with the deployments we have made. More info: http://www.sonicwall.com/us/products/Backup_Recovery.html
Hope this helps!

Advertisements