What follows is a number of key areas that you should think about when considering the move. Hopefully there’ll be at least one of them that makes you think about a topic you hadn’t considered before and results in you making a more informed decision if and when you make the proverbial leap to a dedicated server.
Expense
It should be noted that gaining exclusive rights to a physical server normally comes with significant cost. Typically a host can sell a server many times over to hundreds (or thousands) of clients; selling the same spec server to one client therefore will always be less profitable to the provider, and so the cost relative to the client is much higher. The magnitude of this difference depends on the specification of the server, the environment in which it’s housed and the additional resources that come included with it, for example bandwidth, I.P. addresses and support.
There has been a relatively recent trend towards cheap dedicated servers lately. Personally I think it’s worth bearing in mind the old adage "You get what you pay for". Bargain basement server companies are making cuts SOMEWHERE to cut costs. You need to know where, or if they’ll charge you for "extras" that other companies include.
Performance
In certain businesses, it is imperative that a response is generated within a certain timeframe, or that the system operates even during peak periods. If your site responds slowly would your visitors be adversely affected? Performance is the most common reason for considering the move to a dedicated server yet people consistently overlook the all important detail.
Hardware Specification
How high a spec server are you actually getting for your money? As of going to print (Mid 2009) you should be expecting Dual Quad Core (8 Core) 2.5 GHz CPU’s or greater, at least 4 Gig of RAM and SAS hard drives. Don’t buy legacy stock! Old equipment is a bad place to start, especially if you intend on getting a few years use out of your server.
Hardware Redundancy
If you are moving to dedicated hardware, you want to eliminate as many single points of failure as possible. Look for RAID (Redundant Array of Independent Disks) hard drives; RAID 1 is good for the Operating System (OS) and databases, and RAID 5, 6 or 50 is ideal for files. Dual power inputs, from separate battery backed-up UPS (Uninterruptable Power Supplies) provides continuity in case of a black-out or power spike on one phase.
Network
The most common hardware component to fail is the Network Interface Card (NIC).
Does your server have more than 1?
Are you single or multi-homed to the network, and the Internet?
How over-subscribed is the datacentre’s bandwidth?
A word of warning here; a popular tactic, especially amongst the cheaper hosts, is to provision hardware that isn’t capable of fully realizing its potential. For example, 100Mbps unmetered bandwidth servers often come with a single SATA hard disk or a weka CPU or small amounts of RAM that can’t actually fill the connection!
Flexibility
The power to make your own choices is probably the second most popular reason for choosing to go dedicated; your own server gives you autonomy. If you have specific operating system requirements, kernel tweaks, custom libraries, bespoke software applications or plug-ins it’s likely that a shared server environment will slow down or forbid deployment of these. Be sure to check that you have full root access and can run the server how you want to before committing.
Data Security
Data protection concerns are the third most common reason for dedicated server purchases. How valuable is your data? If you have personal or business sensitive data that would damage your brand or reputation if leaked then a dedicated server offers significant benefits. It’s very rare but servers can be root compromised through another users account. In these cases the hacker can get all the data off the machine. If it’s your server, with only your website on it then you can have all but the essential ports on the machine shut down, restrict access by IP and know that you’re the only one on the server. Some providers will be able to provide you with a dedicated hardware firewall, often for additional cost, which gives you further piece of mind.
Support
Don’t take out a contract with any company that doesn’t offer a full Service Level Agreement (SLA) to cover outages and problem resolution.
An example SLA might include:-
* A technical support response, often guaranteed within a certain timeframe.
* A hardware replacement policy; whose responsibility is it if something breaks and needs replacing? How long will it take?
* Remote reboot, again possibly with a time guarantee.
* Connectivity to both the network and the Internet is often guaranteed in % availability, look for at least "4 nines" (99.99%) uptime, don’t buy into 100% uptime, if something’s too good to be true…
* Look for a company that’s willing to be held accountable for the quality of service they provide, you need to know that they are serious about keeping your server running.
Conclusion
In summary, weigh up carefully whether the need to move to a dedicated server is warranted either on the grounds of performance, flexibility or security (or some or all of these reasons). Naturally it has to make financial sense too, but if it does then it’s highly likely you won’t look back provided you pay close attention to the hardware and company you are committing to. Treat the process like selecting a partner, and you and your server will have many happy days and long nights together!
Source: http://www.articlesnatch.com/
Servers are complex machines - orders of magnitude more complex that the PCs that run at our homes. Apart from the hardware, much of a server’s complexity arises from software interactions. So when you see even companies like Google displaying error messages, you know that something is wrong with the way their software is set up.
Due to the heavy load that servers experience, events that have a minuscule chance of occurring at any given point happen frequently. We have to get used it. What can be done however, is learning how to quickly deal with errors so that your IIS server doesn’t crash.
To this end, IIS 7 comes with sophisticated error tracing tools that allow you to get to the heart of the problem, thereby enabling you to fix it. So what are the common causes of these errors? Here are a few.
Application changes
A new update or bug fix to an existing application can have unexpected effects on the workings of even unrelated software. This is why Server administrators are paid so much. They know exactly what they’re doing and understand the inner guts of a server.
Server Configuration changes
IIS 7 has a lot of options and configurations. Sometimes, you might be forced to make a change that has unexpected ramifications, causing requests to fail.
Why Error Pages aren’t enough
When a web request fails, one can get very detailed error pages that sometimes lead to problems being resolved quickly. However, when the error is caused by interacting modules, error pages simply aren’t enough or are misleading. What can happen is that an error can occur and go unnoticed for a while and when it manifests itself, the real reason is hidden. Debugging is an art, but with the right tools, it can be done much faster. And this is where IIS 7 tracing comes into play.
IIS 7 Tracing
IIS 7 has sophisticated tracing tools that allow administrators to view the worker processes and their requests in real time. The new feature is called Failed Request Tracing and is one of the hallmarks of IIS 7. It’s completely configurable and can be used to trace requests based on certain parameters that are set by the administrator.
For example, and administer can set an alert and tell IIS 7 that if a request takes longer than x seconds to complete, then a trace needs to be initiated that will enable administrators to fix the problem. This command can also be used to fine tune the server and figure out where it is running slow.
If you’re using IIS 7 as a server in any environment including a VPS, you can use the trace features along with IIS 7 remote administration in order to effectively manage your server.
Source: http://www.ezinearticles.com/