Five years from now, will a 200-person company need any servers on site?

Last week I was asked an intriguing question: given the trend towards cloud-based services and applications, is it possible that a 200-person company will soon be able to have no servers onsite? If not, what will the future threshold be?

Last week I was asked an intriguing question: given the trend towards cloud-based services and applications, is it possible that a 200-person company will soon be able to have no servers onsite? If not, what will the future threshold be?

Even in a cloud-centric world of the future, most people would probably agree that large enterprises will never become 100% cloud-based. In other words, large enterprises will probably always have internal datacenters and servers. Most people probably also agree that even today small companies (say, five people) don't need to own any servers—they can get along fine with hosted email and file syncing and online accounting, etc.

So as IT becomes more service-oriented, the point as which a company requires onsite servers (in terms of users) will go up. Today it might be five users. Next year it's ten. Then twenty. So how far will that go? When will we get to the point when no company (regardless of size) needs onsite servers?

And if so, what technologies will drive the servers and apps to the cloud? Or what will keep them onsite?

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

It's already possible. I worked for a company that provided a completely hosted IT infrastructure delivering applications via Citrix and managing resources via a custom-developed web portal. Literally everything is hosted (desktop applications, databases, mail, network storage, etc.) and maintained by the company, including managing disaster recovery. For all of their clients, only a select few needed to have servers remain on site and that was primarily due to manufacturing needs where there are physical serial connections required to run the equipment and companies that were heavy into graphic design / CAD applications. Citrix and the associated network requirements are simply to heavy to easily transition those factors over to the cloud (though, even that is changing).

Things that helped with pushing companies towards this model are:

- multiple office locations and the concurrency issues from having separate file stores

- a desire to focus and have existing IT staff work on projects that are more innovative / benefit the company moving forward than doing day-to-day maintenance

- staff comfortable working entirely online (this also easily enables home offices)

- no significant capital expenditures every 3-4 years for server / desktop refreshes.

I would say that not only is it possible, it's inevitable for companies above 25 employees. The cost benefits far outweigh keeping things as is. The economies of scale also really start to kick in for the MSPs providing the services allowing for a higher level of service for a cheaper price for those companies that make the transition.

My 2 bits anyway.



In 200 years "I think there is a world market for maybe five computers" [on site]. ;)  

Seriously, I do not for a second beleive there wil be that much of huge cataclsymic change in mere 5 years, thus the fun with the reverse figures and uavoided joke with the Thomas Watson 1943 quote


I like the idea of having all servers virtualized somewhere in a cloud, one is tempted to say "it's all a matter of service level", however, I guess it's not.

The company I work for (as external consultant) is good example. They will IMHO never give up hosting and operating the major part of their systems themselves. The reasons for this are:

- regulatory compliance forces the company to qualify lots of systems and processes and keep track of all changes and related data

- company secrets: data from or systems involved in research & development processes are desired to remain self-hosted, no secrets ought to leave the company

- a couple of server & workstation roles demand the systems being placed at special locations to control production (sub)processes or connect to specific laboratory hardware

The strategy by now is to virtualize anything possible and consolidate data centers, anyway I think this will be it for the next decade.



for me the biggest question is: And what happens if the provider where you host all your data  goes bankrupt? You have to make sure you have a really good emergency plan for that scenario, i'm sure most of the small companies don't have these plans and they get into huge problems if the hostingprovider goes bankrupt.


Brian, Great question to ponder. Some thoughts:

1. (@jan Jonker)  Cloud computing (CC) will evolve to be de dominated by a few large primary service providers through whom smaller application vendors will distribute applications/services. This is already the case in mobile telecom - many of the core services we consume over the mobile phone (from a telephony perspective (not IP))  are actually enabled by 3rd part technology vendors - e.g voicemail, MMS, SMS  are functional add-ons provided by 3rd parties.  The service provider turnkeys the whole solution and the consumer doesn't deal with individual service components but consumes and pays for the whole service. (Sequoia backed Bubble Motion is a great example of Voice SMS that is delivered as a carrier service). Cloud apps will be no different.

2. I think in understanding how CC will commoditize, its useful to look at the (E)PABX and telephone industry.  Most EPABX has folded from being on-premise to being service provider delivered (from the telephony cloud).  The same will happen to basic CC services. We already have evidence of this with email services, CRM, FB etc

3.  Whether desktops will become a CC delivered service is upto debate - certain desktop classes are certainly candidate (single task, stateless session oriented desktops, kiosks) but it doesn't fit the vast majority of desktops due to the diverse nature of applications and their maintenance needs.  

5. Realistically 5 years is too short a time ... the Windows internals are too hard wired for the physical world that it was originally written for. Perhaps over the next 2 iterations of windows it starts to look more like an OS that supports JIT assemblies of core OS components that are schedule-able over the WAN.

Chetan Venkatesh


Atlantis Computing


2, 200 or 2000 users...the day there is a major cloud outage, something as simple as your own route to the cloud, for longer then 2 hours, your servers are coming back onsite...


The future is now... Virtual Bridges just partnered with Rackspace to offer VERDE Hosted edition to serve just this market.

Nice timing Brian!


holy vendor spam! that too  on - looks like your akismet plugin needs an update Brian


I'm a network/systems admin for around 130 person company in the financial industry so I would say no.

It depends way more on the business the company is in. For example in my company it is impossible to go 100% SAS. Probably about 1/4 of the company or more are IT people, we run about 50-60 servers (most in VMware) and our own zOS mainframe. For the normal employee they need a few common software, like Office, PDF readers and so on but the most important software is always developed inhouse. This could be somewhat stored on an external cloud, but the risk is far too great.

So, it will depend on what kind of company it is. For large non technical companies I see nothing stopping them from going 100% SAS in 5 years, or even right now. Enterprise Gmail, Salesforce definitely goes a long way.

My vision is still, private clouds with easy management for the majority, public clouds for those very web centric.

We need more easy to get, easy to setup, easy to maintain virtual appliances.


@Chetan -- so, just because Virtual Bridges is offering a hosted SMB solution today for 100, 500 and 1000 user scenarios, that is exactly what Brain is talking about - Rackspace hosting virtual desktops and taking the servers out of the 200-person company - you call that spam? As opposed to your self-serving, vendor-published opinions that the problem is too hard and we all have to wait for you to solve this really hard problem?

This is a very relevant post in that people are free to know that the world is not limited to VMWare, Citrix and their complicated and expensive ecosystems. Real, innovative, comprehensive solutions exist today that address this EXACT scenario and are backed up by world-class hosting companies like Rackspace - and are available at a very affordable price.

Does this solve every possible scenario today, no. But, it is certainly worth the majority of small-medium businesses looking at, and... it will certainly evolve rapidly to where on-site servers are no longer necessary - to Brian's point.

Jim Curtin


Virtual Bridges


As the system admin for a 175 employee non-profit company in the mental health field, I don't believe it would be possible in 5 years time to have a 100% hosted IT infrastructure for our organization.  

If I were to evaluate this concept, I would review the business categories and see where hosted solutions might fit best.

Hosted solutions require a large and reliable Internet pipe.  Most SMB's use a business class cable modem or DSL.  That simply won't do.

VARs (and I used to work for one) usually offer SMBs 'point' solutions and don't really look out for the long term interests of the client.  Very small companies can use point solutions and hosting, but I think as you move up the complexity ladder, in-house IT expertise becomes more important.

For example, as my company expanded to electronic medical records and billing systems, they moved from a VAR to in-house IT staff.  And this was definately the right decision.  Since then, we've improved all aspects of the IT environment, including security, disaster recovery, governance, availability, etc...

The problem with many hosted solutions is just's a solution to solve one problem with little forethought.  SMBs often need the guidence to chose the right solution. I don't see VARs doing a very good job there and the hosting providers certainly won't least from what I've seen thus far.  

They want to sell you on the 'cloud', without ever doing the 'ground work'.

Five years isn't enough time for this evolution, but it will eventually take place.


For companies using Windows OS it's already a reality and by the end of the 2010, 1000s of companies will be 100% cloud based with us. Inhouse they will have nothing but cloud thin clients and a big internet pipes. We are just delivering 250,000 seats for one organisation which will be 100% cloud based.


We uniquely deliver the windows os (XP pro or 7) from the cloud to internet enabled devices at $7-$8 (Sorry to any TS hosts who charge circa $100 per month)

We uniquely deploy any application to our cloud application grid and virtualise every application delivering a SAS solution for any application

whether SAS enabled or not.

So yoru get your windows desktop anywhere to any internet enabled device, monthly licensing model for all your favs, all your bespoke apps and software licenses virtualised.

Voila, cut your IT budget by 80%, whilst greatly improving the standard, redundancy and mobility of your systems.

Nathan Stewart

Cheif Techincal Director

The Portable Office Company


Current cloud provision (like Nathans) is all very good if you have simple use case. The moment you have a diverse user with divers requirements then that blows cloud out of the water. Or a diverse business requirement.

What about Identity? Finance records? customer data? staff data? solitaire, etc? do you really want all those "offsite"?

I don't think the current "cloud" is ready for this yet. Hosting the desktop/desktop session I think is more realistic now but not the rest of the business systems.

However.... 5 years is a long time and I would think for a 200 user business (with relatively straight forward needs) it is more than likely going to be the primary option.

I don't think bandwidth will be a major issues in 5 years (depending on location).

It boils down to the services you offer. Email, web, "some" desktops, fine.... The moment you stick my personal data in the cloud and it gets compromised/lost - I'm coming for ya. - Now or in 5 years time!


@jim - nothing personal here - the BM site and esp. the comment stream are among the most  valuable resources to dialog and move understanding forward.  I do try very hard to avoid any overt/covert self promotion and in this case was making a personal opinion.  Rather than take offense with you for calling me self serving,  I realize I stand corrected by your response. Maybe its not my place to judge if your comment and link to your PR is "spam" given that I too represent a vendor with an agenda in this industry.   Its really up-to the people, the SBC & VDI practitioners  who think about, evaluate, buy and implement  these technologies  to make that judgement.   Thanks for taking the time to respond. Hope no permanent damage was done and we can still be collegial.


Chetan Venkatesh


Atlantis Computing


I'm with ^



We uniquely deliver the windows os (XP pro or 7) from the cloud to internet enabled devices at $7-$8 (Sorry to any TS hosts who charge circa $100 per month)

Duh? and always the latest software ? how is that possible?


@Nathan - I visited and noticed your services are "coming soon" and CANNOT currently be purchased.  Your site also claims it will offer the "first cloud based windows desktop with integrated business communication services".  I love vaporware

On the other hand, Amazon's EC2 ( provides an amazing array of functionality.  Cloud based desktops, servers, block storage, databases, web services, VPN's, two-factor authentication, replication and more.

Citrix has even performed a scalability analysis of XenApp running on the Amazon EC2 cloud (

Does anyone know of a service provider that is currently offering a richer suite of cloud services for infrastructure?  I would like to see a head-to-head comparison of the top tier providers.

I can see the day in the very near future where businesses decide whether or not a system is important enough to house in the "onsite" datacenter or if they will run it in the "virtual" datacenter in the cloud.


Guys. This "Nathan" is a total scam without any legitimacy whatsoever. Check out the marvellous demo (index.php matters...) - I checked it for the fun of it, I wish I haden't...

Can we please have some moderation on the comments?


This all depends, as usual, on the required appplication estate.  The size of the company is completely immaterial.  A hypothetical 200,000 person company that only requires e-mail is already perfectly provided for in cloud services, a one man band that needs custom software is going to be far more difficult to provide for.

Although my experience of smaller companies is limited, I have almost never seen an organisation without any custom applications.  The abillity to host those apps on third party sites is crucial.  Although this can now be done with EC2, the expertise required is greater than that required to set this up in house.