Hosting|Co-location vs managed hosting: News - Business
Co-location vs managed hosting: News - Business - ZDNet Australia
Co-location vs managed hosting
By Josh Mehlman, Technology & Business magazine
January 21, 2002
URL: http://www.zdnet.com.au/news/business/soa/Co-location-vs-managed-hosting/0,139023166,120262977,00.htm
No matter how large or complex your Web hosting requirements, there's someone willing to take them off your hands. But how do you choose?
While hardware sales are slumping and ISPs are finding their services increasingly commoditised, Web hosting providers are looking forward to a profitable future.
Local companies are expected to spend over AU$700 million on Web hosting in 2005, up from AU$106 million in 2000, according to market researcher IDC's report The Australian Web Hosting Market 2000-2005. "That's a compound annual growth rate of 46 percent," says IDC analyst Lisa Shishido.
Why is hosting becoming so popular?
"The biggest reason is companies are moving to e-business, from simple static Web sites to transactional Web sites that are integrated at front and back end," says Shishido.
"There's also a greater acceptance of outsourcing; people are more comfortable with it. Especially as hosting a Web site becomes more complex, companies want to focus on their core competencies. Also, there can be great cost savings from leveraging a hosting provider's economies of scale. It can help improve your time to market." In addition, hosting providers can make it a great deal easier to scale up if your requirements suddenly expand.
However, the market is still very immature. "Companies are still coming up with standard offerings," says Shishido. "For that reason, they can be very cagey about releasing their pricing details."
Co-lo or managed?
The essential difference between co-location and managed hosting is the degree to which the hosting provider takes care of management and administration tasks.
Ben Joseph, Asia Pacific sales director for Intel Online services, says there are several levels of management that can be provided. Bare bones co-location services provide rack space in a data centre, an Internet connection, and will ensure the physical wellbeing of the machines. Everything else is left to the customer.
Managed co-location services provide additional shared infrastructure, such as firewalls, VPNs, load balancing, and staff to perform maintenance tasks. A managed service provider will take care of the entire hosting equation, including management of the operating system, Web server, databases, application servers, and the customer's applications.
"If you keep this in mind, there aren't many true MSPs in Australia, mostly co-location or managed co-location services," says Joseph. Obviously, co-location is more suited to companies who already have technical staff capable of managing the systems, and just require an Internet connection and the space to host their servers.
On the other hand, companies that need to develop a Web presence can outsource to a managed hosting provider without having to hire technical staff. This is one of the reasons there's a strong trend away from co-location to managed hosting.
"Co-location providers are shifting more into managed hosting because it's a way to provide value-added services," says Shishido. "When people look at MSPs, they don't look at them in the way outsourcing is normally seen. People look at them more like a utility or telco. Rather than oursourcing a whole business process, they can maintain some responsibilities where they feel they might have a vested interest or core competency. It's like a utility company where you can pick and choose your services.
"Co-location has a small share of the market compared to dedicated hosting and shared hosting. That's small in terms of customer base and revenue."
Dalibor Vrsalovic, president of Intel Online Services is more blunt. "Co-location is a loss leader, it's always covered by some other part of the business. "Companies that tried to do co-location alone have failed. There's not enough value-add, and the customer still bears the cost of the overheads."
What are the Benefits?
"Hosting without management is only raw connectivity. It's not redundant or secure and it's only a matter of time before it goes down or is compromised, either maliciously or by a worm," says Lorenzo Modesto, sales and marketing director of MSP Bulletproof Networks.
The management that an MSP specialises in can add layers of:
- Security: The MSP can invest in security and firewall equipment that would be prohibitively expensive to a single company. In addition, the technical staff can ensure that OS, antivirus, and application patches are always up to date.
- Redundancy: This again is assisted by the MSP's economies of scale, and might otherwise not be affordable.
- Monitoring: The MSP actively monitors the performance of the machines. If anything goes wrong, or looks like it might, the MSP's specialist engineers can be paged quickly.
- Reporting: The MSP can provide detailed reports of the systems' performance, which can be used for forward planning of requirements.
"All of which turns raw connectivity into a platform for mission critical applications including e-commerce applications," says Modesto.
Most importantly, managed hosting can save money. "Cost comparisons are very clear cut when you add up staff costs, stress, lost revenue, and lost customer satisfaction if a site is not hosted on a managed platform."
"Managed Hosting eliminates or greatly reduces all of these so that SMEs that rely on Web Hosting can get on with using it as the business tool it should be."
How to choose
"We recommend that they look at a company that is financially secure, employs skilled staff and comes across as knowing what they're doing," says Modesto. "Contact with the staff needs to be smooth and professional."This might all seem like pretty obvious stuff, but it's still surprising how rarely you find these attributes on hosting and co-location companies."
Modesto thinks the seven most important attributes for a hosting provider are:
- Redundancy: There should never be a single point of failure. Make sure their internal and external network connections, power supplies, backup power systems and all other systems have redundancy--if one ever fails, another should be available to take up the slack.
- Connectivity: Speed of connection to the outside world is an important factor, but a diversity of links to different carriers ensures faster performance for end users no matter which ISP they are using. Australian sites looking to attract local customers must have strong connections to local carriers.
- Network management: Networking staff should have the skills and management tools necessary to handle any problems.
- Availability and escalation procedures: How easy is it to get hold of someone in an emergency? How does the provider contact you if something goes wrong?
- Monitoring: All systems should be proactively monitored on a 24x7 basis. Multiple forms of monitoring and performance metrics are ideal.
- Security: Security procedures, both for physical access to systems and for protection of data, must be solid. Yet they shouldn't be so inflexible that you can't get access to your systems at 3am on a Sunday, if need be.
- Service Level Agreements: While all hosting providers offer SLAs (or if they don't look elsewhere), there are several important issues in drafting an SLA to keep in mind, according to Modesto.
- Clarity: They should be written in plain English, not technical or legal jargon.
- Simplicity: The parameters for measurement and the claim mechanisms should be specified in the SLA, and should be stated as succinctly as possible.
- Fairness: Do the sums, and make sure the conditions under which you can claim result in a reasonable rebate. This should be based on a multiple of total downtime. Pihana Pacific general manager Doug Oates adds that while most SLAs will guarantee five nines availability (the systems are available 99.99999 percent of the time, which equates to about five minutes of downtime per year), the SLA should also set out who measures this availability and how. "Also, do the penalties have any teeth? Is the facility designed so it can realistically provide that service? Just looking at the SLA is one thing, but also at the design and infrastructure behind that."
Bandwidth trading
Because some large data centres have links to multiple carriers, some are now also setting themselves up as bandwidth exchanges, allowing customers to trade bandwidth on different networks."We operate an Internet exchange service that allows customers to exchange traffic with each other within the facility," says Pihana Pacific's Oates.
"We encourage the use of that because we're not in the bandwidth business. We see that is a more and more important value add because it minimises the need to buy bandwidth from your upstream provider. We'll be facilitating peering and exchange of traffic."
This not only allows customers to gain access to several providers, improving the consistency and reliability of their sites' performance, but can also allow them to gain access to extra bandwidth in times of high demand.
Automatic for the people
Managing Web hosting systems is becoming increasingly complicated, and around 52 percent of all problems are caused by human error, either accidental or procedural, according to Intel's Vrsalovic. "If we learn from history, then it's obvious this must be automated."
Like the telephone exchange and the automatic teller, Vrsalovic believes machines will be able to take over the mundane and repetitive tasks system administrators currently have to do.
Intel has developed a controller system for its managed hosting facilities, that uses a rule-based engine with artificial intelligence to run scripts that mimic the behaviour of a human operator.
"The controller addresses the issue of the enormous people cost involved in keeping these systems running," he says. It also makes administrators' lives easier. "Operators are under enormous pressure when something goes wrong, and this can create even more problems. Imagine instead having a button to press that says 'fail over'."
Copyright © 2008 CNET Networks, Inc. All Rights Reserved.
ZDNET is a registered service mark of CNET Networks, Inc. ZDNET Logo is a service mark of CNET NETWORKS, Inc.
- Redundancy: There should never be a single point of failure. Make sure their internal and external network connections, power supplies, backup power systems and all other systems have redundancy--if one ever fails, another should be available to take up the slack.
2 comments:
Thanks for the concise explanation!
Web Hosting Services
Webcare360 Provides you safe and secure cheap Offshore Hosting Best offshore hosting and offshore Host By Webcare360 with 99.9% Up time Guarantee, with Ddos protection.
Post a Comment