Difference between revisions of "Hostek infrastructure"

From Hostek.com Wiki
Jump to: navigation, search
(Server virtualization with high availability)
Line 40: Line 40:
 
*All servers are connected to the storage network via multiple paths for performance and redundancy
 
*All servers are connected to the storage network via multiple paths for performance and redundancy
  
Additionally, all data on our storage arrays are replicated to standby storage devices daily.  This means that should an array cluster fail, the replica can quickly be brought online to resume storage.
+
Additionally, all data on our storage arrays are replicated to standby storage devices daily.  This means that should an array cluster fail, the replica can quickly be brought online to resume storage.  See figure 2 below for an example of normal operations.  Figure 3 is an example of a storage array failover situation.  Note that in the failover situation, "Data volume 1" and "Data volume 2" become available on the standby array and virtual machines depending on those volumes may resume normal operations.
  
 +
 +
[[File:san.png|link=|figure 2]]
 +
'''figure 2'''
 +
 +
 +
[[File:san-failover.png|link=|figure 3]]
 +
'''figure 3'''
  
 
==IT security==
 
==IT security==

Revision as of 20:52, 25 June 2012

A significant amount of effort and expense goes into designing and maintaining the network infrastructure to ensure maximum performance and availability of services hosted by hostek.com. Below are some of the highlights of the hostek.com infrastructure.


Bandwidth providers

Hostek.com utilizes multiple Internet bandwidth providers. By utilizing multiple vendors for Internet bandwidth, service to the hostek.com network avoids interruption from vendor maintenance and outages. See figure 1 below.


Internet bandwidth providers figure 1


Server and storage hardware

Hostek.com is proud to utilize Dell servers and Dell EqualLogic storage arrays.


Server virtualization with high availability

Our web and email servers are virtualized utilizing VMWare vSphere. With VMWare High Availability, should a physical server fail, any virtual machine will be automatically migrated to another physical server and powered on, and resume normal operations. This operation can be completed with the affected virtual machines back in service often in 3-5 minutes! This is an amazingly short time frame to recover from a failed physical server.

Another added benefit to virtualization is that resources such as CPU, RAM, and disk space, can be quickly allocated based on utilization patterns. This ensures optimal performance for your site or mail server with minimal down time.

Due to the nature of certain kinds of workloads, some servers may not be virtualized.

Redundant power

Electricity is is provided from two separate power feeds. In addition, our equipment is protected with UPS systems and diesel generators. Servers and storage devices have redundant power supplies, each on separate feeds to help prevent loss of data due to power failures.


Storage

Virtual machines, along with their data (i.e. your web site files, email, etc.) reside on enterprise storage arrays. These arrays have the following features:

  • Multiple arrays clustered for high performance
  • Redundant storage controllers
  • Redundant hard drives
  • Redundant network adapters
  • Redundant storage network switches
  • All servers are connected to the storage network via multiple paths for performance and redundancy

Additionally, all data on our storage arrays are replicated to standby storage devices daily. This means that should an array cluster fail, the replica can quickly be brought online to resume storage. See figure 2 below for an example of normal operations. Figure 3 is an example of a storage array failover situation. Note that in the failover situation, "Data volume 1" and "Data volume 2" become available on the standby array and virtual machines depending on those volumes may resume normal operations.


figure 2 figure 2


figure 3 figure 3

IT security

Firewall & IPS

A permemiter firewall and Intrusion Prevention System (IPS) exist which perform the following functions:

  • Stateful packet inspection
  • Deep packet inspection for known malicious attack patterns
  • DDoS (Distributed Denial of Service) protection mechanisms

SPAM filtering

All inbound email is filtered using multiple real-time RBL and content inspection technologies.


Additionally, outbound mail from our Windows web servers are checked with anti-SPAM content filters to prevent delivery of mail from your site being disrupted by compromised mail submission forms from other tenants on the server your site resides on.


Anti-virus

Our servers run commercial anti-virus products from reputable software vendors.


Backups

We perform daily backups of all web, email, and database servers.

Backups for VPSs are stored off-site in a location that is in a different geographical area than the primary site. This means that should a major disaster occur in the area where the servers running your VPS reside, your data is safe in another location.


Physical security

Restricted access to premises

The buildings housing our datacenter have a comprehensive security system which include but not limited to the following:

  • 24x7 on-site security
  • Access codes
  • Biometric hand scanners
  • Electronic proximity readers
  • Security surveillance system


Fire suppression system

Our equipment is protected by a pre-action, dry pipe fire suppression system.


PCI Compliance

Click here for details on hostek.com's PCI Compliance.