As both are designed for storing and retrieving information, hard drives and Wikis are very similar in some regards. Both have read/write statistics, though it is much easier to measure the read and write speeds of a hard drive. When buying hard drives, one can look at the speeds before purchasing. Comparing the random and sequential read and write speeds between drives is relatively simple. Even though measuring methods may differ between manufacturers, they can still be used to get a general idea of how a drive will perform. Additionally, it is possible to buy several drives and evaluate their performance on your own test. This will give you results that you can compare between drives while knowing that they have been obtained in the same matter. With a Wiki, it is almost impossible to measure read and write speeds. They are heavily dependent on the person accessing or storing the information: different people work at different speeds. The write speed is the more variable and person dependent; it is affected by too many variables to reliably improve. The read speed, however, can be vastly improved through various methods.

Our chosen method for decreasing the time it takes to retrieve information from our Wiki is to standardize the pages. All pages that cover similar topics, such as customer contact information, will look the same: all the information will be stored in the same manner and in the same place. This improves read time by allowing technicians who access the information to know where to look. Were all the pages subject to their own formatting, precious time would be wasted attempting to locate the information, even after locating the page on which it is stored. This can increase the downtime for a server which can cost you money. As we strive to offer the best experience possible, this scenario is unacceptable: your hardware should never be down because our documentation is confusing. Our Wiki is organized to make it as easy as possible to retrieve information, but there is still the possibility for failure.

The MTBF for both hard drives and a Wiki indicates the time between being unable to read information from the system. Whereas with hard drives this can be corrupt data or simply a broken drive, with a Wiki this tends to be more along the lines of a user who is unable to locate the information necessary or, in the worst case scenario, information that is simply missing. Both of these failures can cause service interruptions and, in a worst case scenario, one can even delay or prevent remedying the other. As a company dedicated to providing the best service possible, we find both of these scenarios unacceptable. We believe that you should never be inconvenienced because we omitted or misplaced information. To help prevent this, we have taken steps to simplify and facilitate the process of adding information to our Wiki.

The standards mentioned above play an important role in reducing, and hopefully eliminating, misplaced information. In our Wiki, we believe in a place for everything and everything in its place. However, there is still the human factor. People will make mistakes while entering the data. Through proofreading and spell-check, we can attempt to minimize these occurrences, but they will still happen over a long enough period. The restrictions imposed by the standards pose another risk: how do we prevent people from abandoning the standards when creating new pages? Even though I designed the standards we put in place, I would likely begin to slip if I had to manually format each new page. Typing the formatting syntax is tedious. To combat this tedium, we have made some changes to the way the Wiki looks for page templates. This has allowed us to use a separate template for each page type. Our “Contact Info” page, for example, can have its own template distinct from our “Server Info” page, allowing us to pre-populate newly created pages with the proper formatting. By pre-populating the pages with formatting, we remove much of the tedium associated with documenting our customers’ information.

Though Wikis and hard drives serve a similar purpose–storing information–and present similar problems–read speeds and reliability–they require distinct solutions. These solutions, however disparate, allow us to overcome some of the shortcomings of informational storage. To increase the read speed of a hard drive, you might purchase a higher-RPM drive or a solid state drive. To increase read speeds on a Wiki, we have organized our data uniformly, decreasing the amount of time spent searching for information. To mitigate the risk of drive failure, you might set up RAID or have a backup server on standby. For our Wiki, we have removed as much of the tedium of data entry as possible, allowing our employees to focus on what is important: storing information in a manner that allows us to serve you, our customers, as best as possible.

Selecting the Perfect Site

It all started with scouting out the perfect site. We got lucky and were able to find an excellent location near our old HQ in the high-demand Herndon, Virginia area. The Hallmark building was ideal for our needs with the infrastructure it had in place and its proximity to major highways and Dulles International Airport.

Move-in

After our site was selected, work began on getting walls painted, furniture and data center gear moved in and systems installed. We officially moved into the office portion of Dulles Exchange in Q1 of this year–an exciting milestone for us InfoRelay folks!

Being in charge of the branding for the new facility, I took particular interest to its appearance. Check out the first glimpse I got of our new reception signage thanks to the gracious folks at Impact Signs:

Who knew indoor signs were shipped like that? Another move-in milestone was when the data center cabinets arrived. This was a huge process and took numerous man hours to unload, inspect and install all of them!

Data Center Infrastructure

In the next few weeks that followed, I was witness to several important milestones, including the laying of the dark fiber cable to our building:

as well as the delivery and installation of our generator infrastructure:

The generator delivery and installation in particular was an excellent scene to witness. Just imagine, 28,000 pounds being lifted over your head and moved to its protective containment. They made it look so easy!

Final Touches

After weeks and weeks of hard work, the final touches are being made and the data center officially opened its doors to customers on May 1, 2012. Here’s a good shot of our revamped reception area:

And another shot of the rows of cabinets on the data center floor after installation:

I’m so proud of how far InfoRelay has come, with Dulles Exchange being a testament to our successes and hard work. Next step, to celebrate with a launch party! More details to follow.

If your still hungry for more, I have plenty of pictures up on the Dulles Exchange website as well as its technical stats and how to schedule a tour. Come see it for yourself!

Herndon, VA–InfoRelay Online Systems, Inc., a leading provider of on-demand data center and cloud hosting services today announced that their latest data center, titled Dulles Exchange, is now open to customers.

Located within the prestigious Hallmark Building in Herndon, Virginia, the new facility will cater to medium and large enterprises that require a commitment to in-depth technical support and mission-critical power, cooling, and network connectivity. The new location will also be home to InfoRelay’s headquarters and will help the company to accommodate their recent growth and future plans for expansion.

Founded in 1995, InfoRelay has been profitable since its inception and continues to experience high levels of demand for its colocation and cloud offerings. “Based upon the incredible demand that we’ve seen, especially in Northern Virginia, combined with our desire to ensure that customers have the highest levels of redundancy available, it made perfect sense for us to design and construct our own facility. By doing so, we’ve bolstered our position in the market as a full-service data center service provider and opened up the doors to a myriad of opportunities for our customers as well as our company.”

Like all of InfoRelay’s data center locations, the Dulles Exchange data center is highly secure, diversely connected to multiple fiber networks, and features highly redundant critical infrastructure systems. Biometric access controls, 24×7 staff, site surveillance, state of the art flywheel UPS systems (n+1 and 2n availability), diesel generator power, automated controls, and fully redundant self-healing network connectivity are just some of the standard amenities offered at the Dulles Exchange data center.

A broad range of services will be available at the new data center, including InfoRelay’s renowned 24x7x365 customer support, colocation and dedicated servers, high-speed bandwidth and connectivity options, data backup and recovery, and private and public cloud services.

Dulles Exchange will be ready for customers by May of 2012. The official grand opening is scheduled for mid-June, 2012. Guests or clients interested interested in learning more about the service, or who are interested in attending the launch event can contact sales@inforelay.com for more information or to schedule a tour.

Flash Me No More!

Flash-based content has permeated websites in droves for years, and it’s proven its value time and again with enabling various multimedia experiences that are either difficult or impossible to emulate with the existing open Web standards. Some sites utilize Flash seamlessly with standard W3C-compliant content, while others have created their entire site completely with Flash, which has allowed for some pretty spiffy animated and engaging sites that showcased the future of the web before the official standards caught up with it. In fact, Flash site generators made it even easier to get started on a site—an attractive option for those who cannot afford a good web designer for their site or who just want to get up and running, pronto. This is of course not to mention the success of such Flash-heavy sites as Youtube, which has—newsflash—taken the world by storm, garnered the attention of a major search engine company enough to be bought out for almost $2 billion, and has spawned countless clones.

This all may have made sense years ago; Flash seemed to be the future. However, with spotty support on devices in the ever-growing gadget market, and with advances in recent years of both the HTML and CSS standards, as well as some creativity on the part of website designers, Flash has become increasingly obsolete. In its place advances the more open and community-backed official standards, which have become modernized to meet the demands of today’s consumers, who want to experience the same content across their multiple devices—home computer, mobile phone, tablet, etc.—without having to worry whether and how well their device supports Flash (looking at you, iOS devices). Although the newest versions of HTML and CSS aren’t quite as ubiquitous as Flash, they are set to be more compatible with a broader range of devices. There are already some online games that leverage HTML5 and other technologies to offer unique experiences. BrowserQuest is a good example: it plays similarly, regardless of which HTML5-enabled browser you use. My first time playing that, I got killed by a rat that was admittedly half my size

Perhaps in admission of defeat—Adobe itself has publicly announced its resignation of Flash on mobile platforms, and has intended to work on improving HTML5, as it is no longer feasible for them to continue development on a product that is definitely showing its age in the face of newer technologies. Youtube has even been developing an HTML-based alternative. In the grand history of the Web, Flash will have been just a flash in the pan.

My Browser is Better Than Yours ;p

Browser dependencies are more a thing of the past (from the Great Browser Wars of way, way back), but in this age of an open Web, open standards, and cross-browser sites, some sites still advertise that they work best with a particular browser. For instance, my online tool of choice for taxes this year made it clear to me that my browser, Chromium under GNU/Linux, was not supported. I did, nonetheless, have the option to continue, and I found that the stern warning was more of a relic of times past, as I had not a hitch doing my taxes. In fact, if I had used a user agent spoofer so that my browser would masquerade as Internet Explorer 6.0 running on Windows XP, I’m sure the site wouldn’t have known any better, even though it would’ve been the same browser by any other name.

Aside from upfront compatibility messages, there are other sites out there that proudly display badges demonstrating their support for one or another browser (and implicit scorn for all others). For instance, some sites are best viewed in Internet Explorer, while others are too cool for Internet Explorer, and yet others specifically advocate for all browsers. Any sites you see these days that extol the virtues of Browser X are most likely just poking fun at the dark ages when websites really took browsers seriously (or just haven’t yet been featured in Website Makeover: That’s So Yesterday Edition). Chances are there are still quite a few out that actually might not work at all for certain browsers due to proprietary markup. Thankfully, most major organizations and companies have recognized that in order to ensure the majority of their audience is able to view their content without embarrassing browser issues, they must support as many browsers as possible (i.e., adhere to the W3C standards as much as possible).

What About You?

The above are just a couple of the artifacts (or soon-to-be artifacts) you may still encounter during your daily Web surfing sessions, as I have. Over the course of the Web’s growth, there have been countless blunders with certain website features that somehow were made more common than they really should have been (a feat that I’m sure will continue to be repeated in the future). I encourage you to sound off in the comments below with any in particular that may have struck a chord with you and is either on its way out or is still woefully too common. It doesn’t even have to be about a certain technology, like Flash, it could be just a really grating trend you’ve noticed during your InterWeb travels. What do you say?

Herndon, VA — InfoRelay Online Systems, Inc., an IT solutions and connectivity provider, announced today that it will be launching a new “Service as a Service” (SaaS) platform.

InfoRelay, a leading provider of managed colocation, high-bandwidth connectivity, and dedicated server hosting has been serving the public and private sectors since 1995. Since its inception, InfoRelay has focused heavily on providing high levels of service to its clients. InfoRelay has also been a pioneer in using virtualization technologies to operate more efficiently, ultimately conserving power by fully using all of the resources of each server, eliminating the need to run excess underutilized equipment, improving ROI (return on investment) and helping the company to approach carbon-neutral status. Such technologies have already been used by InfoRelay to power its cloud platform. The company believes that its new Service as a Service platform has a plethora of uses, including social media marketing, microblogging, neuromarketing, trending, solutions-based viral marketing, personal branding, as well as niche technology applications, such as mobile marketing, Geolocation, and even NFC (Near Field Communication).

“The convergence between real world social interaction and online interaction has truly been a game changer,” states an InfoRelay Executive. “When you look at the gamification that is often used to create brand loyalty, a prime example of which is FourSquare, it’s only natural that companies would want to pull the trigger and go after the low-hanging fruit by leveraging a value-driven, solutions-based, customer-centric turnkey-solution. It’s difficult to argue with the fact that online is The New offline. There’s really a paradigm shift in America; I took off from work the other day and I was telling everyone that I’d be out of pocket. The truth of the matter is that I had never been more in-pocket. At the end of the day, aside from having to wipe away some pocket lint, it was a win-win situation for my company and for me – all thanks to my OLED smartphone and some thinking outside of the box. It’s a knowledge economy, and companies are going to be right-sizing when they begin to understand the consumerization of intellectual capital and the importance of the 80/20 rule and the associated metrics. I think the synergies that we’ll see between content marketing, personalization, and brand partnering will truly push the envelope, reinforcing not only the globalization, but the localization of today’s economy. We see InfoRelay’s virtualized cloud-based Service as a Service Platform as the starting point, and something that will lead to the ultimate tipping point in the realignment of content marketing in order to increase social engagement. ”

Through open-sourcing, InfoRelay believes that the potential for third-party value-adds to its SaaS platform in today’s highly competitive marketplace are virtually unlimited. Excitement currently abounds from all members of the InfoRelay staff; one member had this to say:

“We are really hitting the ground running, taking it to the next level, and giving it 110%. No, scratch that, my calculations were off… make that 120.816%. Truthfully, no other company in the industry has built its core competency in the way that we have. I asked one of our competitors about their core competency in an open meeting – they wouldn’t answer, and just responded with ‘Why don’t we take this offline?’. As they say, it is what it is, and what it is… is maximizing customer satisfaction through seamless integration, scalability, and green initiatives. This is a net-net win-win, an integrated approach to moving up the value chain. For us, SaaS is the 800 pound gorilla that brings a lot of value to the table. We offer a best-of-breed solution focused not on just the bottom line, but the bleeding edge. I feel that due to our SaaS offerings, we can now truly open our kimonos and show our clients our 800 pound gorilla – we will not drop the ball.”

InfoRelay has stated that it has begun to architect several additional solutions that will be based upon its Service as a Service offering. Such solutions and innovative offerings include SPaSTIC (Software Platform as Service Tied Into Cloud), CB-RadSEO (Cloud-Based Radical Search Engine Optimization), SaaSQUATCH (Service as a Service Query-Unified Application Turnkey Cloud Hypervisor), BAD SPEW (Bandwidth and Deduplication Service Platform Engineering Widget), and DooDo (Distributed Object-Oriented Data Offshoring). Ultimately, the company states that its end-goal is to increase green initiatives and by Q4 2015, with planned capital investment for R&D of nearly $10 million, to launch a fully green platform known as SaaSaFRaS (Service as a Service Application Firewalled Rapid Acceleration System). SaaSaFRaS will be in industry first in that it will actually be carbon negative, meaning that it will consume carbon dioxide and produce oxygen. The launch of this service will likely coincide with significant improvements to virtualization technologies, resulting in a second-generation (2G) version of SaaS, known as SaaSaaS (Service as a Service as a Service). The company expressed gratitude for the progress they have made, and therefore the ability to announce at the beginning of Q2 2012 (April 1), the overarching initiative responsible for all recent technology developments; FaT E-BUTT (Fun at The Expense (of) Buzzwords Used Too Typically).

Note: This article is purely hypothetical and in good fun in honor of April Fool’s Day (if you can’t tell).

In this case, the customer in question was able to quench the attack by identifying and blackholing the target of the attack, and then moving the resources to another address. This is usually much easier than trying to determine the addresses of all of the attackers. Since most attacks of this type are automated, the odds were with our customer that the attackers wouldn’t refocus on the new address. In this case, they were correct in doing so. Blackholing their own IP address proved to be the easiest way to stop the attack from affecting their network.

Cutting off Your Nose to Spite Your Face?¹

It may be counter-intuitive to blackhole your own IP addresses. After all, you’re the victim of the attack, not the perpetrator(s)! What you should keep in mind is the trade-off between one or a few customers being affected by the attack, and all of your customers being affected by the attack. If the attack traffic can saturate all of your connections² to the Internet, then your entire network is down. It’s usually better to inconvenience a small number of customers than to inconvenience all of them.

How Does This Work?

InfoRelay Online Systems protects our network with various equipment and techniques for blackholing this traffic. Some of these methods are available on demand by our BGP customers. One of these methods involves using a special BGP community to alert the InfoRelay network that a certain address needs to be blackholed. This method involves three steps:

• BGP customers advertise smaller IPv4 prefixes to the InfoRelay network tagged with our blackhole community.
• The InfoRelay peer detects that prefix, sets next-hop to a null route, and shares that prefix with all other InfoRelay routers through iBGP.
• Finally, every router that has an eBGP peer with an applicable blackhole community adds the original prefix to that provider’s blackhole community, so that our eBGP peers will also null route this traffic. In cases where we peer directly with an eBGP blackhole router, we also send the newly-tagged prefix to that router, so that the traffic is blackholed.

Advertising the Target

The first step is triggered by the customer. Through any number of means³, the customer is alerted to the attack, and determines the destination address on their network. Once the destination of the attack has been identified, the customer sets up a special route announcement consisting of the destination address or addresses. InfoRelay can blackhole any IPv4 or IPv6⁴ prefix up to a /32 or a /128, respectively⁵. The customer must then ‘tag’ the route with the InfoRelay blackhole BGP community. That community is 33597:666⁶.

Automatically Blackholing Traffic

The second step happens automatically, without intervention by the InfoRelay NOC. Once an InfoRelay router detects a new route tagged with the InfoRelay blackhole BGP community, it takes the following actions:

• The router adds the new prefix announcement to any existing prefixes in the InfoRelay blackhole BGP community.
• The router alters the prefix announcement by setting ‘next-hop’ to a null-routed interface. This prevents all IP traffic from the InfoRelay router in question from reaching the target of the attack.
• The router then advertises this new route to the InfoRelay network. Through iBGP, all InfoRelay routers learn about the prefix to be blackholed. They in turn set ’next-hop’ to a null-routed interface. This prevents all IP traffic bound for the target from traversing the InfoRelay network.

Alerting Upstream Providers

As mentioned in the introduction, we’re seeing larger attacks all the time. In order to prevent these attacks from saturating our connections to our upstream providers, we piggyback on their blackhole capabilities and ask their network to also blackhole the targeted address. We do this by two means:

1. We ‘tag’ the new route with additional upstream provider blackhole communities and send it to all upstream provider networks, thus triggering their blackhole capabilities.
2. We send the new route over BGP sessions with upstream providers’ blackhole routers. This in turn also triggers the providers’ blackhole capabilities.

Both of these techniques result in the same outcome: IP traffic bound for the target is blackholed by our upstream providers, and never reaches the InfoRelay network, much less the customer network. At this point, the attack should be quenched, and the customer can start moving targeted resources to alternate addresses, in order to restore services.

Summary

DDOS attacks happen all the time, often for no apparent reason. Odds are against you receiving any sort of warning prior to an attack. InfoRelay Online Systems and other providers leverage BGP communities to provide automatic means of suppressing traffic bound for addresses under attack. This allows customers to turn these capabilities on and off at their desire, without the hassle of a support ticket escalation. This in turn allows their networks to remain responsive under great stress with a minimal impact. If you’re not currently configured to use these capabilities, you should consider doing so.

1 See http://www.phrases.org.uk/meanings/106875.html for an explanation of this phrase.

2 You’re not single-homed, are you?

3 This is worthy of several articles on its own. It’s outside the scope of this article.

4 It was easy enough to add IPv6 support, so we did so, in an attempt to ‘future-proof’ our capabilities. I don’t know of any IPv6 DDOS attacks on our network, but they will be here soon enough.

5 One major caveat is that the prefix in question must fall under a prefix that’s already being accepted by the InfoRelay peer. This community does not allow you to blackhole arbitrary addresses. If the attack traffic is coming from a small number of identifiable sources, then it’s better handled through a support ticket and an ACL provisioned by our NOC. Also, there’s nothing preventing a customer from blackholing their entire address space. That would be as bad as it sounds. Welcome to the big leagues.

6 I don’t know if this is from an RFC, but many networks use the Number of the Beast for their blackhole community. InfoRelay is not an exception to that rule.

With their less-than-reliable lifespan, it is important to honor the sounds of the past. So please, crank up the volume and prepare to relive the memories of our technology ancestors that the 20-somethings of today may have never experienced. Courtesy of IT World.

There are a number of other cloud services out there that you might hear about. In many of these cases, the term cloud is being used loosely to mean stored on our servers in our data centers. For example, cloud backup means that you backup to your provider’s servers in whatever data center they are in. Google Docs and Fickr are examples of other cloud-based solutions where you are storing your data on someone else’s servers.

Both of these are examples of software-as-a-service (SaaS). There are many famous examples of this approach. The majority of SaaS is provided through a web browser to someone with an Internet connection. There are some exceptions such as Microsft Windows licensing or RedHat licensing when used for hosting. SaaS most often seems to appeal to end users but it can be used anywhere in an organization. SaaS very often uses IaaS to back it up.

Finally, the most virtualized cloud technology is Platform as a Service (PaaS). Remember that to provide IaaS, you have divorced the Operating System (OS) from the server and have divorced the storage from the server. Software glue makes sure that the OS keeps running smoothly as it is moved around. With PaaS, we further divorce the software from the OS.

To provide PaaS, a provider must offer a programming language, target environment and a storage layer. This software stack is divorced from the underlying OS. If designed properly, the PaaS platform is able to dynamically expand to handle very large workloads. As is true anytime one is dealing with high traffic, high capacity services, software developers must be careful in their designs to avoid bottleneck. This sort of design involves making tradeoffs. PaaS providers choose some tradeoffs upfront. You may give up the ability to use any language and platform you want as well as guarantee of short term data consistency. In return, you are given the ability to scale greatly and have assurances that your data layer will remain durable and reliable.

Virtualization has come to influence many of the activities we perform on the Internet. There are several different virtualization technologies that all work together to provide some amazing possibilities. These technologies are as good as the data centers, people and software/hardware stacks backing them up. InfoRelay’s unique industry position can help you to make the most of virtualization in your organization.

There is a lot of information to learn regarding these topics. I hope this article has helped you to get a solid understanding of virtualization technologies and how they may be useful to you and your organization.

Herndon, VA–InfoRelay Online Systems, Inc., a leading provider of on-demand data center and cloud hosting services, and PC Recycler, a nationally-recognized Federal compliant data destruction and electronics recycling firm, today announced a partnership to deliver a holistic solution catering to the complete needs of their broad customer base.

The strategic partnership will deliver best-of-breed services from both companies. Serving businesses and federal agencies with a focus in the Northern Virginia area, GSA contract holder PC Recycler will provide compliant destruction and recycling services using industry-leading techniques. The service promises zero landfill disposal and zero export policy. With its robust service portfolio, Herndon-based InfoRelay provides a wide range of enterprise-level managed IT services from highly-redundant server colocation, bandwidth, DDoS protection, to automated cloud-based desktop and server-grade backup solutions.

The new alliance between InfoRelay and DoD compliant PC Recycler seeks to add value to customers by offering a more complete set of services, utilizing the strengths of both companies. The program will allow InfoRelay’s clients, which include Federal and commercial organizations, to dispose of hardware, drives, and other electronic peripherals on-site at select InfoRelay data center locations. PC Recycler will perform the asset destruction on-site to ensure that data cannot be lost in transport.

“Forming a partnership with PC Recycler seemed as though it would add a lot of value for our clients; many of whom have sensitive information stored on their equipment. As our customers cycle through equipment or replace failed hard drives, they need to ensure that sensitive data is destroyed. Our partnership with PC Recycler allows us to do just that,” stated Russell Weiss, President of InfoRelay.

InfoRelay will be deploying a hard-drive destruction program for use by its customers in select facilities and PC Recycler will be offering data center services through InfoRelay by Q3 of 2012. Customers interested in hosting services or data destruction services can contact sales@pcrecycler.com or sales@inforelay.com for more information.

# # #

About InfoRelay Online Systems, Inc.

InfoRelay Online Systems, Inc. delivers enterprise-level managed services and wholesale data center services to thousands of clients, including Fortune 500 companies and federal agencies. InfoRelay improves its clients’ efficiency and profitability through the company’s array of reliable IT services. By demonstrating unparalleled levels of responsiveness, concern, and overall service, InfoRelay establishes a unique trusting relationship with each client, allowing InfoRelay to act as a 24x7x365 extension of its clients’ IT departments. For more information, please visit www.inforelay.com, follow InfoRelay on Facebook and Twitter, or call (703) 485-4600.

About PC Recycler

PC Recycler is an emerging leader in information technology waste management and data destruction. With its headquarters is in Chantilly, VA, PC Recycler has been a provider of recycling and data destruction services to large and small corporations, municipalities, and private institutions for five years. PC Recycler has taken a different approach from most waste management companies by utilizing an innovative cost effective formula in helping companies with the disposal of their electronic waste and complies with DoD, NSA, NIST, NISP, HIPAA, FACTA, PCI, and a host of other requirements. For more information, visit www.pcrecycler.net.

The idea of virtual servers started long ago on IBM Mainframe computers. In the last decade or so, virtual servers have come to PC-class hardware. Virtual servers work by taking a host computer and running other operating systems under a host operating system. This works because there is virtualization software that “tricks” the guest operating system by trapping certain requests and emulating the hardware response. The last few generations of processors from Intel and AMD have provided special instructions for making this process more efficient.

These guest operating systems were generally called virtual servers for a few years. They were usually sold on a single physical server and were mostly used as a way to give someone some of the benefits of a dedicated physical server such as root/administrator access without giving them their own server. Companies quickly saw the benefit of using this technology to consolidate systems. It is possible to run many different servers under their own virtual machines on the same hardware. The major limitation to this was that in many cases, early generation virtual servers were relying on the host computer’s local disk.

Soon, virtual server providers started to realize the benefit of using virtualized storage for the disks of virtual servers. This lead to the Infrastructure as a Service (IaaS) cloud computing model. Clouds are clusters of physical servers connected to virtualized storage that allow for the easy move a virtual machine from one physical server to another and automatic restart of a virtual server if it fails. A simple expansion capacity by starting new virtual servers quickly is what people often mean by “elastic computing.”

One of the primary differences between cloud computing and other services is the billing model. Cloud computing often used utility based billing. Rather than paying for a service each month on a yearly contract, some companies are now providing people with the option to pay a each month for the number of hours each virtual server was running. If your workload varies greatly, this can be a big help. Be careful however, the rates are generally higher than for a similar dedicated server. The utility billing model can also be complex because you have to pay for the server size (generally tied to RAM), bandwidth in, bandwidth out, storage in use and disk operations. This complexity can make something that sounds cheap like $0.40/hour end up being much more costly than they seem.

Next week – Part 3: Other Cloud Services.