Virtual private servers (VPS) – have historically offered some middle ground between shared web hosting services and dedicated hosting services, both in control and cost, but recently the gap in cost between virtual and dedicated servers has diminished … making upgrades from shared to dedicated web hosting very attractive.

A virtual private server separates you from – other customers on a physical web server, running on a copy of its own operating system, but it shares CPU and RAM resources of the physical web server with other VPSs running on that server.

A dedicated server, on the other hand, is – a physical web server leased and controlled by the end client, without sharing physical CPU and RAM resources with other sites. Its very much like owning your own server, without the huge capital asset investment.

Most hosts offer a mix of unmanaged versus managed services for dedicated servers. Unmanaged plans typically offer the basics – the hardware, operating system, control panel and Internet connection.

In a VPS, a single physical server is partitioned – so that it appears as multiple servers. The physical server boots normally, then runs a program to boot each VPS within a virtualization environment.

There are two kinds of virtualizations – software and hardware based. Software based environments share the same kernel and require the main node’s resources. In a web hosting environment, quota incrementing and decrementing in real time is possible without restarting the node. In a hardware based virtualization, real hardware resources are partitioned, eliminating burst or real time quota modification. This lends to a (potentially) more secure environment.

With a dedicated server, you have – more flexibility and control. Dedicated servers have historically been the server of choice for complex business or high end eCommerce sites, but have recently become very popular substitutes for VPS servers and less complex solutions.

Dedicated servers housed in data centers offer – redundant power sources, HVAC sytems, state of the art security and advanced performance services.

The bottom line: Moving websites from a shared environment to either a VPS or dedicated server is a significant event. Its a major upgrade in the IT infrastructure supporting your business. Whereas the move from shared to dedicated was historically cost-prohibitive, and VPS offered some middle ground, technology has rapidly evolved to narrow that gap.

Upgrading should be based on more than cost

Upgrading your IT technology, in this case to VPS or dedicated is now a business decision based on more than cost. It’s based on resources you control that will map your business success for years or decades.

With the price gap narrowing, dedicated services are rapidly becoming the plan of choice for business savvy entreprenuers.

Linux is distributed in - different flavors, and like cars, which flavor to use depends on the perception of the user. So what are the underlying reasons to select one versus another? On the surface, Ubuntu, Debian, Fedora, openSUSE, CentOS and others look pretty much the same.

All releases contain - more or less the same set of basic packages, and at the kernel level (with the same version installed), they’re all basically the same – Linux.

There are subtle differences - between distributions though, like variations in their file system hierarchy or layouts. Other variations differ in terms of system behavior, initialization procedures and anatomy of their scripts & names.

The differences are – generally very superficial and in the end, Linux is essentially nothing more than the kernel.

If you’re looking to use cPanel – then I’d recommend CentOS. I found an excellent tutorial this morning on Setting up CentOS, and this from CentOS.org:

CentOS is an Enterprise-class Linux Distribution – derived from sources freely provided to the public by a prominent North American Enterprise Linux vendor. CentOS conforms fully with the upstream vendors redistribution policy and aims to be 100% binary compatible. (CentOS mainly changes packages to remove upstream vendor branding and artwork.) CentOS is free.

It seems this question keeps popping up on forums across the Internet. Maybe this helps …

A virtual private server (VPS) – separates you from other customers on a physical web server, running on a copy of its own operating system, but it shares CPU and RAM resources of the physical web server with other VPSs running on that server.

A dedicated server – on the other hand, is a physical web server leased and controlled by the end client, without sharing physical CPU and RAM resources with other sites. Its very much like owning your own server, without the huge capital asset investment.

Most hosts offer – a mix of unmanaged versus managed services for dedicated servers. Unmanaged plans typically offer the basics – the hardware, operating system, control panel and Internet connection.

In a VPS - a single physical server is partitioned so that it appears as multiple servers. The physical server boots normally, then runs a program to boot each VPS within a virtualization environment.

With a dedicated server - you have more flexibility and control. Dedicated servers have historically been the server of choice for complex business or high end eCommerce sites, but have recently become very popular substitutes for VPS servers and less complex solutions.

Atom dedicated servers are – essentially solution driven appliances, as opposed to more powerful (and expensive) Xeon-based servers. Their energy efficiency integrates well with any number of applications, including storage and network appliances & print and email servers.

If you don’t need extreme computing power – atoms are smaller, less expensive and more power-efficient microprocessors that deliver just enough power to get the job done. You may be surprised that Atom processors are found in everything from smartphones to netbooks.

Servers are normally built for – maximum load, but for the majority of time, they sit idle, at least from the perspective of the CPU (unless you’re running virtualization). During low traffic, high-end Xeon driven dedicated servers are not especially power efficient.

When Intel introduced their low-power Atom chipset – a couple of years back, they were limited by the amount of RAM they could max out (2GB). With the release of the Atom D510, that jumped to 4GB.

So, what can you do with an Atom dedicated server?

• Run Apache, PHP and MySQL applications – putting your own apps on top of your favorite operating system
  • CentOS 5 – 64 bit
  • Fedora Core 13 – 64 bit
  • Debian 5.0 (Lenny) – 64 bit
  • Ubuntu Server Edition – 64 bit
  • openSUSE 11.3 – 64 bit
  • Windows 2008 Standard – 64 bit
• Run multiple blogs
• Run a weather station
• Run as a file server
• Run your own telephone switch
• Run a firewall

What hardware configurations affect dedicated servers across the board?

Processor speed, cache and CPU Front Side Bus (FSB) are important. Size, speed, type of hard drives and RAID configuration are factors. The amount of RAM is significant as well. But what about the speed of that RAM? How important is it, really?

In comparing identical servers, certainly one with DDR-200 Mhz RAM will underperform today’s standard of 667Mhz DDR2 RAM. Slower memory speeds relate to diminished server performance. Be cautious when spec’ing out your server – even though the FSB may be 1333Mhz, the speed of the RAM may be significantly less. If the RAM speed isn’t listed in a prospective providers offer, simply ask.

When is RAM speed necessarily relevant?

You’ll primarily see a difference in very memory intensive operations. For most, the difference in speed won’t be noticeable. RAM speed accounts for only a few percentage points difference one way or another. In general, speeds limited due to RAM are more likely because you don’t have sufficient RAM, and not because the RAM is too slow.

What happens when your server pages out to disk?

Quite simply, memory runs in nanoseconds and hard drives in milliseconds – no comparison. Once your server starts paging out to disk, server performance will slow dramatically. Consequently, if you’re seeing speeds limited due to RAM it’s more likely you don’t have enough RAM, not because the RAM is too slow. Unless you’re a gamer or benchmark analyst, you probably won’t see much difference between 800Mhz or 1333Mhz RAM.

RAM Explained

ECC (Error Correction Code) versus non-ECC RAM 

ECC RAM will typically run a couple of percentage points slower in performance, but its advantage is increased error handling capability and system resilience. ECC will recover from single bit errors having additional ‘bits’ in order to detect and then correct any ‘parity’ errors in memory. In designing and building servers, select ECC RAM for improved stability.

Peak RAM Speed 

The number (400/533/667/800/1066) represents the raw peak transfer capacity of the memory chips used on the module. Higher speeds have the potential to move data faster, but often that’s accompanied by higher latencies. Frequent small memory access may be better served by lower latency RAM. Frequent large block transfers may be better served by higher peak transfer speeds (despite an increase in latency). 

Unbuffered versus Registered and Fully Buffered 

Unbuffered RAM will generally offer a slight performance boost over Registered or Fully Buffered. Registered and FB adds electrical buffering to the pathway which allows for more RAM to be added to a system but at the cost of increased latency. 

Registered RAM

Registered RAM has buffer chips that increase the amount of RAM a computer can have. You can never use registered memory in motherboards that require unbuffered memory or vice versa.

Both are generally found in server-type hardware, but unbuffered ECC works in desktop/gaming hardware.

Fully Buffered RAM

There’s also “fully buffered” RAM (“FBDIMM” or “FB-DIMM”) which is similar to registered RAM, but unlike normal registered RAM which only buffers the control and address lines, FBDIMMs also buffer the data lines. Similar to ECC (Error Checking and Correcting) memory, FB-DIMM modules perform error checking and correction as well as cyclic redundancy checks on data passing through them, thanks to their AMB chip. The AMBs in each memory channel co-ordinate their error checking efforts so that if a correction or retry is necessary, all the modules stay in sync.

Capacity is where FB-DIMM really does seem to have a clear-cut advantage over conventional memory. Where previously there was a hard 4GB memory limit, modern 64-bit processors can potentially address effectively unlimited amounts of memory at full speed.

Here is a breakup of what you need where:
- AMD socket 754: unbuffered DDR, ECC or non-ECC
- AMD socket 939: unbuffered DDR, ECC or non-ECC
- AMD socket 940: registered DDR, ECC
- AMD socket AM2: unbuffered DDR2, ECC or non-ECC
- AMD socket F: registered DDR2, ECC
- Intel socket 478: unbuffered DDR, ECC or non-ECC
- Intel socket 775 (Core2Duo etc.): unbuffered DDR2, ECC or non-ECC
- Intel Xeon Woodcrest (51×0) and Dempsey (50×0) socket 771: FBDIMM DDR2, ECC
- Previous generation Xeon socket 603/604: registered, can be DDR or DDR2 (depends on chipset/board), ECC
 
With AMD versus Intel CPUs 

In the AMD world, the CPU socket type determines whether you need registered or unbuffered RAM. In the Intel world, it is the chipset and hence the motherboard. That’s why some socket 603/604 boards require registered and some require unbuffered. Existing socket 771 chipsets all require FBDIMM.

Memory Interleaving 

This will typically increase overall performance a couple of percentage points. Interleaved memory allows the system to fetch concurrent banks of data from multiple DIMMs simultaneously. This helps to compensate for the latencies of main memory.

Dedicated hosting offers customers the option to lease pre-configured servers and Internet connectivity for their exclusive use versus VPS or Shared, where other customers share the host server.

Why lease a dedicated server?

For smaller customers, dedicated makes more business sense.

A broad range of managed services are available.

The server belongs to the hosting provider and it’s their responsibility to maintain and keep it running.

Entry costs are lower and scalability is enhanced.

It allows you to outsource IT expertise so you can do what you do best – drive your business

If you’re on a VPS or shared account now, some considerations to move to a dedicated solution may be:

Traffic to your site has increased to where it’s approaching the limit allowed on your current plan.

Your marketing projections reflect dynamic growth, drastically accelerating timetables to upgrade.

You’re seeing reduced bandwidth speeds, possibly because your current host is overselling their servers.

Obviously, the return on your investment (ROI) must be justified, both in hard and soft dollars. Dedicated servers will cost more than VPS or Shared. Can you manage your own server or will you need managed services?

The majority of entry level and mid-range servers are configured with 120v plugs even though by design they will work with 120 or 208v. Larger towers and rack mounted servers tend to be configured with 208v.

Why 208v?

The primary reason is for power supply energy efficiency, which is a measurement of the amount of power lost between the input and output of a server’s power supply. You could expect a two to three percent increase in power supply efficiency by simply increasing the input voltage, resulting in a significant saving in power expense. Some have reported as high as a 30 percent increase in efficiency.

This energy savings is significant to any data center’s bottomline.

Can 120v be retrofitted to 208v? Absolutely. Whereas 120v has one hot, one neutral and one ground, 208v simply has 2 hot lines versus one.  Do your servers run 120v or 208v? Will your prospective data center customize their existing 120v to match your 208v requirements? These are important questions to ask.

Why is 208v more energy efficient?

For a given amount of power, as the voltage increases, less current is required. Conversely, as voltage decreases, more current is necessary to maintain the same amount of power (what we pay for).