Author: Yogi Schulz
Many organizations struggle with the decision to outsource the hosting of their Web infrastructure. It’s tempting and seemingly cost-effective to simply expand the existing in-house data center to accommodate the new Web-accessible applications.
However, the overriding objective for most organizations is to achieve consistent high availability for their Web-accessible applications. For many organizations, the availability target for their in-house data centers is much lower, making these facilities unsuitable for Web-accessible applications.
Here are some criteria on which to base the decision to host or not to host. For most organizations, the result of applying these criteria to their own circumstances is to outsource hosting.
Access to the Internet Backbone
The data center of a hosting vendor provides redundant high-capacity access to the Internet backbone. This capacity translates into an always-on, snappy surfing experience for your audience.
The cost of this capacity is difficult for a single organization to justify. As a result, the audience of an in-house hosted Web application is more likely to experience reduced performance and more outages.
Hosting vendors hire sufficient staff to operate the servers around the clock. This staffing is affordable because it’s shared across many clients and is particularly useful to reduce problem resolution times.
With this staff, the hosting organization can monitor the health of servers and databases continuously. When signs of trouble appear, corrective action can be taken quickly.
Most in-house data centers are staffed only during daylight hours. Sometimes the skill or experience of the staff does not meet the needs of the web-accessible applications. The cost and boredom make around the clock staff untenable.
Hosting vendors have dedicated staff to execute security tasks such as check-in and check-out of personnel, managing security cards and monitoring the Web infrastructure. Most valuable of these tasks is pro-active monitoring of firewalls and server logs looking for intrusion indicators.
Security operation is not performed with rigor and consistency in many in-house data centers due to budget and skill constraints.
Hosting vendors have developed and exercised disaster recovery plans because it’s part of their service commitment to their customers. Many have developed relationships with other hosting vendors to provide fail-over capacity in the event of an actual disaster.
For many in-house data centers, the disaster recovery plan has been deferred for many years under the relentless pressure to reduce operating costs.
Physical Access Control
Many hosting vendors have implemented a combination of security card validation and biometric identification for physical access control.
Such rigorous physical access control is too expensive to acquire and operate for most in-house data centers.
Data Center Location
The Web infrastructure is best located in a remote building with limited people traffic. The advantages include lower cost, easier-to-recognize intrusion attempts, superior security and access control and improved environmental stability.
In contrast, most organizations locate their in-house data center within their office building. While convenient, this approach makes adequate security and access control more difficult due to the large number of people in the immediate vicinity. In addition, the ongoing demands of tenants for renovations create electrical power and environmental outages that adversely affect the data center.
To minimize the impact of electrical power disturbances, dual power sources, significant battery capacity and standby diesel-powered generators will protect the Web infrastructure.
Not only are these components costly for most in-house data centers, they are often defeated by events such as planned power outages and unexpected fire alarms in typical office buildings which house the in-house data center.
The Web infrastructure needs a consistent temperature and humidity to achieve maximum availability.
Most office buildings don’t achieve the desired level of consistency and offer no redundancy. While this is fine to meet the needs of office dwellers and the demands of cost accountants, it’s insufficient for Web-accessible applications.
The use of the higher capability Web infrastructure described here, to host Web-accessible applications, will cost more than use of the in-house data center.
The justification is that the cost increment is necessary to maintain the organization’s reputation for quality and reliability through highly available and performant Web-accessible applications. It’s an investment that’s needed to ensure that the benefit case for the Web-accessible applications is achieved.
The central point is that an organization’s Web infrastructure usually requires a higher level of operational sophistication and a higher level of facility investment than many organizations are able to justify for the balance of their computing infrastructure.
Web infrastructure, to function well on the high-visibility Web, should be outsourced to a capable hosting vendor.