Sunday, August 15, 2010
The Divorce and Going Foward
I apologize to any of you who read my blog and were awaiting further posts. The fact of the matter is that Wyse and the Wyse Guy have agreed on a parting of ways. As you can see from my previous posts, my goal is to create a hardware agnostic client environment which enables any of a number of different client virtualization methodologies. The Wyse sales leadership finds that this approach would not be an enhancer to Wyse hardware sales, but rather a competitive offering which would be a detriment. As you can see from the Wyse website, the company focus is now Cloud Client Computing (CCC) which is really oriented around Desktop Cloud Computing and Desktop as a Service (DaaS) virtualization offerings. While I believe that both SaaS and DaaS are the most efficient forms of client computing, the infrastructure to provide these offerings is initially expensive and the applications rarely translate perfectly to these environments (look at the difference between MS Office and Office Web Apps). As the world awaits these environments' build out, the client hardware technology is already being developed and put in the user's mainstream portfolio. Netbooks, Tablet PCs, SmartPhones and other UMPC devices are becoming more and more commonplace both at home and in the office. Thus, software solutions which make these devices more secure and enable these devices to be multi-purposes are the order of the day. Wyse offers both streaming technology and PC repurposement technology to make better use of these upcoming technologies and make virtualization more of a mainstream solution. That said, do not expect the Wyse sales team or partners to sell you these solutions. While Tarkan Maner (current Wyse CEO) is definitely a proponent of this movement, none of his executive team back him up with more than words. If a customer specifically asks for one of the non-hardware solutions, they will first be pursuaded that they need a full hardware solution and then be told that they are required to purchase certain minimum amounts of software even on a trial basis. I have seen license opportunities on software pushed away or just completely ignored by both the local sales rep and the VP of Sales. I stepped in where possible, but even I had to concentrate on the larger opportunities. Thus I gave up on this executive team as a lost cause. That said, I have not given up on thin computing. Thus, I am looking for a new moniker under which to pen my posts. Anyone reading this should let me know what they believe is an appropriate name to laud thin computing, desktop cloud computing and DaaS.
Friday, April 9, 2010
PC Extender, The Future of Thin Clients?
Recently, Wyse made a splash with Project Borg. The product has since been released and is now called PC Extender. The premise of this product is that any existing PC can be transformed into a thin client without changing the hardware and theoretically without touching the device itself. PC Extender is an Microsoft Windows Installer program which requires 1 GB of free space on the hard drive of the user. What it proceeds to do when run is to take back that 1GB of free space and create a bootable partition on which the Wyse Enhanced Suse Linux Enterprise for Thin Clients is installed. In addition, a boot manager is installed which is set to automatically point to the newly created Wyse partition to boot the device. When the device is rebooted, it reboots as a Wyse thin client. This thin client can be managed exactly as if it were a Wyse thin client using the same management tools. Additionally, all of the Wyse thin client extension (TCX) components which are available under Linux can be used on this device including Wyse Multi-monitor, Multimedia, USB Virtualization, Rich Sound and Flash support. Also, the Wyse Virtual Desktop Accelerator can be used to increase the usability of the protocol in high latency or error prone networks.
The real value of this product can really only be realized when one looks at the use case scenarios in which the product really plays (please see my other posting for why a thin client is almost always better than a re-purposed PC for the enterprise):
The real value of this product can really only be realized when one looks at the use case scenarios in which the product really plays (please see my other posting for why a thin client is almost always better than a re-purposed PC for the enterprise):
- Providing a bridge for users with an existing PC environment which is still being amortized to gain the value of a thin computing environment without the capital expense of thin clients
Not everyone is immediately ready to throw out their PCs. In fact, many still have either a lease or an amortization investment for these devices. This investment is often a deterrent for even thinking about virtualization because of the cost of maintaining both environments. Thus, a major transformation is put off until a large chunk of devices are ready to come off the books. PC Extender allows for the investment to be made with only a minor investment of the physical client side. Additionally, it provides a fallback scenario as it does not erase the existing hard drive, but rather creates a new partition and leaves the existing installation in place. Thus in the case of a major network catastrophe where the users cannot reach the backend environment, the users could work with non-network applications on the other partition by just re-booting the device. - Adding other groups to an existing pilot/proof of concept projects without having to upset the apple cart
Often, Pilots and Proofs of Concept are localized to a single group with a singular set of requirements. As the breadth of the scope is increased, other users must mak the same investment in order to prove their own concept. PC Extender provides a mechanism for these users to be included in the proof of concept without having to make the capital investment in thin clients. This inclusion allows for a more robust environment which encompasses the needs of the entire enterprise versus the requirements of a single group within the enterprise. - Supporting a bring your own computer (BYOC) initiative in an enterpriseA concept which is gaining more and more momentum in the industry is the BYOC concept for user devices. The problem for an IT manager is how to support all of these environments because they do not have the resources to have each type of hardware which is being used. If instead, the device is converted into a thin client for work purposes and converted back into a personal device after work, the user has the best of all worlds and the IT manager does not have the headaches. While PC Extender's install program is only set up for a windows environment, the concept is easy to implement in a MAC environment. MAC environments allow for a dual boot scenario to be created and if a 1GB partition is made available, the Wyse Enhanced SLETC can be installed into that partition thus allowing the device to be a dual purpose device without the headaches of a dual purpose device.
- Providing a secure mechanism for work at home users to attach to the environment
Much the same as the BYOC scenario, the work at home scenario allows a personal device from the uer to be re-purposed to be used as an enterprise device. Many call centers are moving toward a work at home scenario for their employees and PC Extender provides the perfect vehicle for leveraging the user's computing equipment for enterprise purposes while still allowing the user to return the device to home usage when not connected to the enterprise environment.
- Manageability
For the first two scenarios, both the PC environment and the virtual environment must be managed with everything that a Windows environment requires for at least patch management, but also application management. Plus, this risk exists that the virtual environment will not be used completely because the user has access to all of the local applications.
For the second two scenarios, the IT manager has no way of knowing what device will have to be configured to be allowed access to the enterprise environment. Plus, the IT manager must ensure that the correct remote protocol client is available for all possible environments.
With PC Extender, the manageability of the device becomes the same as the manageability of a thin client, which is to say a simple, centralized configuration environment. All clients are managed in the exact same way with the exact same firmware, meaning that the job of the IT manager is how to distribute the PC Extender code. - Security
The problem with any PC environment is that it can be compromised just by the user using the environment. A PC is set up to cache information and maintain that information across multiple boots. A thin computing environment is just the opposite. The thin computing environment was created to connect to a backend environment. Thus the only information which need be cached is that which is required to make the connection to the backend. And this information is so minute that it can be retrieved on te boot of the thin client. Thus, the thin client caches very little locally and in fact anything that is downloaded is purged upon reboot. - Disaster Recovery/Business Continuity
In a typical distributed computing environment, each and every user would have to have multiple stations on which they could login in order to connect to the remote environment. While this is possible in a typical distributed computing environment, each one of the users would have to log into the host environment in order to get their access because of security requirements. By transforming the devices into thin devices, any user can walk up to any device and attach to the remote computing environment. - Productivity
By having a controlled environment for connecting to the virtual environment, the IT manager can virtually assure that the availability of the computing environment is equivalent to the network availability which is typically 99.9999%. Additionally, the virtual machines of the users can be left in an "ON" state so that when the user attaches to their computer, a login sequence is not always required. The user may attach and detach from their running environment regardless of where the user is sitting at the moment.
PCs vs. Blade PCs/Workstations vs. Thin Computing
I have been working in the computing industry for many years now (more than I care to admit). I worked on the revolution from mainframe computing to personal computing. I worked on the revolution from personal computing to web computing. I believe that I am now working on the next revolution in computing, namely thin computing. Many people were surprised by my move from heading up all of infrastructure and security architecture at Deutsche Bank to heading up innovation at Wyse Technology. My answer to my critics is that I felt that the next revolution in computing needs to take place and take place now.
If we look at all of the computing power available worldwide, the number is staggering. There are approximately 1 billion computers in use in both business and home settings (some serving a dual purpose). If we were to make an examination of a single company's end user computing infrastructure, we would find a top utilization of 25% of the CPU. Of that 25% utilization, 90% of the utilization comes from about 10% of the user community. That means that we are leaving huge amounts of processing power to waste away. When you further analyze the fact that that computing power is being exchanged or updated at a rate of once every 3.5 years due to hardware issues and once every 2 years due to technological need, the surprise turns to rage. And when you further factor that on average each of those device is using 200 Watts of power or more (without the monitor), that rage turns to depression.
I did an analysis for a bank which showed that just by exchanging their aging PCs for medium sized (not the super small) thin clients, that they could save $250 per person per year. But I digress to topics for other posts.
There are some simple facts which no person or company can escape today:
As momentum grew for this solution (Citrix is now a 1 billion dollar company because of this architecture), more and more companies came out with competing architectures. ClearCube introduced a blade PC solution for moving the computing power into the telephony closet (and in some cases back into the data center). HP introduced their CCI architecture. Both of these concepts (and some other lesser known concepts including the IBM HC10) are based on maintaing the current relationship between a single user and a compute device, but reducing the size of the device at the end user. This is equivalent to putting a network KVM (Keyboard, Video and Mouse) switch in place for the user. The problem with this architecture is that it maintains many of the issues of the standard PC model while introducing the network display protocol as a new issue.
Finally, a number of different products have been introduced for virtualizing an 80x86 hardware platform. Most notably among them is VMWare. VMWare just this year announced that they would support PCs running as virtual machines in their ESX server infrastructure. This news was well taken because VMWare requires virtually no change to the running PC infrastructure in order to run the in the virtual machine infrastructure. Again, low cost thin clients can be used to connect to this infrastructure. However, this infrastructure also introduces its own new set of issues (again to be discussed in a later post).
If every architecture has its own set of issues, why then do I believe that we are on the cusp of a revolution. The answer is simple: money. While people will usually pay more for the comfort of something that they know, when they are confronted with the savings potential of moving away from their current computing model, I am of the opinion that they will run to a new computing model much the same as when PCs revolutionized the computing industry.
In future posts, I will cover the different thin computing models (including some exciting new models which may just push this technology trend into the technology mainstream).
If we look at all of the computing power available worldwide, the number is staggering. There are approximately 1 billion computers in use in both business and home settings (some serving a dual purpose). If we were to make an examination of a single company's end user computing infrastructure, we would find a top utilization of 25% of the CPU. Of that 25% utilization, 90% of the utilization comes from about 10% of the user community. That means that we are leaving huge amounts of processing power to waste away. When you further analyze the fact that that computing power is being exchanged or updated at a rate of once every 3.5 years due to hardware issues and once every 2 years due to technological need, the surprise turns to rage. And when you further factor that on average each of those device is using 200 Watts of power or more (without the monitor), that rage turns to depression.
I did an analysis for a bank which showed that just by exchanging their aging PCs for medium sized (not the super small) thin clients, that they could save $250 per person per year. But I digress to topics for other posts.
There are some simple facts which no person or company can escape today:
- It is difficult at best to keep the latest and greatest operating system and software on the end user device
- A majority of support calls begin as follows:
I have a problem with my XYZ
What did you do?
Nothing different than I normally do
These calls usually lead to multiple levels of support becoming involved to debug a problem which probably was caused by a software update which occured in the background (either intentional or malicious) - If the problem cannot be solved on the telephone, either a service call is made (read very expensive) or the end user lives with the state of the device
- The cost of adding new end user devices or moving or updating existing end user devices is extremely expensive because users are tied to there device in that they store information relevant to what they are doing on the device
- Computers are by their nature targets for theft and as such, the information that is stored on the device is a target for theft
- As stated above, the environmental cost (this also includes the fact that a computer should be kept in a climate controlled, clean environment which again costs money) of computing is horrendous considering how much of the device's potential we use
- Supplemental hardware costs (local information servers, local print servers)
- Networking costs (bandwidth is required to move information to and from users, i.e. when an e-mail with a 1 MB attachment [small these days] is sent to 50 people, 50 copies of the attachment are pushed to 50 devices in the network)
- Insurance costs (because of the targeted nature of computing devices for theft, companies must insure themselves against that threat)
- Security costs (while this could be perceived as a cost for everyone, the end user usually scrapes by with whatever was sent with the device, but a company cannot)
- Productivity loss (everytime that an end user for a company must take the time to debug an issue on his/her device, that is time which they could have used to do their job)
- Disaster Recovery/Business Continuity (companies must insure themselves against the natural and unnatural threats to their computing infrastructure and ensure that the business can continue)
As momentum grew for this solution (Citrix is now a 1 billion dollar company because of this architecture), more and more companies came out with competing architectures. ClearCube introduced a blade PC solution for moving the computing power into the telephony closet (and in some cases back into the data center). HP introduced their CCI architecture. Both of these concepts (and some other lesser known concepts including the IBM HC10) are based on maintaing the current relationship between a single user and a compute device, but reducing the size of the device at the end user. This is equivalent to putting a network KVM (Keyboard, Video and Mouse) switch in place for the user. The problem with this architecture is that it maintains many of the issues of the standard PC model while introducing the network display protocol as a new issue.
Finally, a number of different products have been introduced for virtualizing an 80x86 hardware platform. Most notably among them is VMWare. VMWare just this year announced that they would support PCs running as virtual machines in their ESX server infrastructure. This news was well taken because VMWare requires virtually no change to the running PC infrastructure in order to run the in the virtual machine infrastructure. Again, low cost thin clients can be used to connect to this infrastructure. However, this infrastructure also introduces its own new set of issues (again to be discussed in a later post).
If every architecture has its own set of issues, why then do I believe that we are on the cusp of a revolution. The answer is simple: money. While people will usually pay more for the comfort of something that they know, when they are confronted with the savings potential of moving away from their current computing model, I am of the opinion that they will run to a new computing model much the same as when PCs revolutionized the computing industry.
In future posts, I will cover the different thin computing models (including some exciting new models which may just push this technology trend into the technology mainstream).
Friday, October 9, 2009
Wyse Does Streaming?
Yes, Wyse has an alternative to standard terminal services and virtual desktop solutions. It is the most feature rich streaming product available on the market, Wyse WSM. The problem is that the Wyse Sales force has difficulty promoting this product because of the relationships with Citrix, VMWare and even to an extent Microsoft.
Today, the Thin Computing market is dominated by the server providers from Microsoft, Citrix and VMWare with a smattering of other offerings such as Parallels. These solutions are the most optimal solutions on the market for enterprise sized customers. However, they require some scale in order to meet the lofty expectations of 25-50% operating expense reductions in a timeframe which would allow an IT executive to keep his job. The other assumptions that these solutions make is that the infrastructure is in place in the data center and the network to be able to handle the transition from a physical environment to a virtual environment. This may change as public clouds become more developed, but it is not the case today for many companies. Additionally, companies must assess the impact if there is a catastrophic network outage or data center failure.
This is where streaming comes into play. Streaming is really an old concept called network booting, but taken to a new level. What occurs is that a small boot disk is booted on a client which has ample RAM, but no local storage. The boot disk can either reside on a USB stick, a small local flash or on a network boot platform which pushes the disk to the client on network boot. The boot disk registers the client with a local server (should be either in the LAN or very close to the LAN) and the server begins to stream across a company specific image of the operating system of choice (in the case of WSM, this is a form of Windows [NT, XP, Vista or 7]). Enough of the operating system is sent across so that the Gina can run and the user logs into the system. It is the user's profile which determines what comes next. The applications to which the user has access are placed into two categories: subscribed and unsubscribed. Applications marked as subscribed are placed on the desktop, in the start menu and/or in the quick launcher. These applications are now available to be streamed when needed. Applications which are marked as unsubscribed are placed in a menu to which the user has access for subscription.
While all of this sounds extremely difficult, it is actually not. To obtain an operating system which can be streamed to clients, the administrator need only suck that OS off of an existing client using a simple tool from Wyse. To create applications to be streamed, the administrator need only use a snapshot tool from Wyse to look at the difference in a system before an application was installed and after it was installed and instantiated. These components are stored one time for all users. In fact, Wyse has developed a tool which will allow a single operating system image to be used for different types of computing hardware. This means that one OS image can serve the entire company.
But the beauty of the product does not lie in its simplicity. Instead, it lies with a couple of major features. The first feature is how changes to the OS are handled. As each of you know, every time you run an application or attach to the web, your operating system files are changed, even if ever so slightly. Malicious items (and some COTS software) intentionally make changes to the OS in order to allow them to do their job. WSM allows the administrator to determine exactly what will happen to those changes. WSM stores a cache file on the local server which is providing for the individual client. This cache file contains the modifications to the base OS image that the user has made. The administrator can choose to wipe out all changes to the OS every time the system is booted (this is called Shared mode). Thus, the cache file is only there temporarily and can be reset by just re-booting the box. The administrator may choose to allow the user to maintain changes that they make to the system (the assumption here is that the system is not shared between multiple users). This Persistent mode allows changes which are not being stored in directory services to continue to persist on this machine, but not between machines. The adminstrator also has a mode in which changes can be made to the operating system and then stored back to the golden image. The administrator mode modifies the image directly and then makes it available to be pushed out to all users either as a patch or a completely new image on the next reboot of the physical end devices.
For those technical people among you, Wyse has also thought about fault tolerance. The clients can be configured with primary and secondary servers (which of course have to store the cache files on some sort of network or RAID shared storage in order to handle all the clients). The network streaming servers can be tiered in order to reduce the WAN traffic. The users can be configured in Active Directory or in LDAP. The applications can be configured to be automatically distributed for all users or all users in a particular group (that means that the application will be mandatory for the user so that they may not unsubscribe from it). Applications can be associated with a domain group to determine which users have access to it. Each application can be configured with an associated license key. The administrator can determine what to do when the license is invalidated (i.e. the 6th user when the license only allows 5 users). The event can be ignored, just logged or the user can be prevented from subscribing to the application until the license has been updated.
The end result of all of these features is a company full of devices which have no local storage, but have a completely secure PC environment without the headaches of managing the PC environment. Support cases are at worst a reboot of the device which results in bringing the user back to a known state.Application development is much easier because only one environment need be tested and the environment can be easily reproduced when issues with an application arise. Deployment is as simple as creating the application (or application update) on the central server and having that server perform a distribution to all network servers. License monitoring and conformity can be ensured by the administrator using WSM. The overall cost per user can be drastrically reduced, especially if the end clients are replaced by thin clients.
Today, the Thin Computing market is dominated by the server providers from Microsoft, Citrix and VMWare with a smattering of other offerings such as Parallels. These solutions are the most optimal solutions on the market for enterprise sized customers. However, they require some scale in order to meet the lofty expectations of 25-50% operating expense reductions in a timeframe which would allow an IT executive to keep his job. The other assumptions that these solutions make is that the infrastructure is in place in the data center and the network to be able to handle the transition from a physical environment to a virtual environment. This may change as public clouds become more developed, but it is not the case today for many companies. Additionally, companies must assess the impact if there is a catastrophic network outage or data center failure.
This is where streaming comes into play. Streaming is really an old concept called network booting, but taken to a new level. What occurs is that a small boot disk is booted on a client which has ample RAM, but no local storage. The boot disk can either reside on a USB stick, a small local flash or on a network boot platform which pushes the disk to the client on network boot. The boot disk registers the client with a local server (should be either in the LAN or very close to the LAN) and the server begins to stream across a company specific image of the operating system of choice (in the case of WSM, this is a form of Windows [NT, XP, Vista or 7]). Enough of the operating system is sent across so that the Gina can run and the user logs into the system. It is the user's profile which determines what comes next. The applications to which the user has access are placed into two categories: subscribed and unsubscribed. Applications marked as subscribed are placed on the desktop, in the start menu and/or in the quick launcher. These applications are now available to be streamed when needed. Applications which are marked as unsubscribed are placed in a menu to which the user has access for subscription.
While all of this sounds extremely difficult, it is actually not. To obtain an operating system which can be streamed to clients, the administrator need only suck that OS off of an existing client using a simple tool from Wyse. To create applications to be streamed, the administrator need only use a snapshot tool from Wyse to look at the difference in a system before an application was installed and after it was installed and instantiated. These components are stored one time for all users. In fact, Wyse has developed a tool which will allow a single operating system image to be used for different types of computing hardware. This means that one OS image can serve the entire company.
But the beauty of the product does not lie in its simplicity. Instead, it lies with a couple of major features. The first feature is how changes to the OS are handled. As each of you know, every time you run an application or attach to the web, your operating system files are changed, even if ever so slightly. Malicious items (and some COTS software) intentionally make changes to the OS in order to allow them to do their job. WSM allows the administrator to determine exactly what will happen to those changes. WSM stores a cache file on the local server which is providing for the individual client. This cache file contains the modifications to the base OS image that the user has made. The administrator can choose to wipe out all changes to the OS every time the system is booted (this is called Shared mode). Thus, the cache file is only there temporarily and can be reset by just re-booting the box. The administrator may choose to allow the user to maintain changes that they make to the system (the assumption here is that the system is not shared between multiple users). This Persistent mode allows changes which are not being stored in directory services to continue to persist on this machine, but not between machines. The adminstrator also has a mode in which changes can be made to the operating system and then stored back to the golden image. The administrator mode modifies the image directly and then makes it available to be pushed out to all users either as a patch or a completely new image on the next reboot of the physical end devices.
For those technical people among you, Wyse has also thought about fault tolerance. The clients can be configured with primary and secondary servers (which of course have to store the cache files on some sort of network or RAID shared storage in order to handle all the clients). The network streaming servers can be tiered in order to reduce the WAN traffic. The users can be configured in Active Directory or in LDAP. The applications can be configured to be automatically distributed for all users or all users in a particular group (that means that the application will be mandatory for the user so that they may not unsubscribe from it). Applications can be associated with a domain group to determine which users have access to it. Each application can be configured with an associated license key. The administrator can determine what to do when the license is invalidated (i.e. the 6th user when the license only allows 5 users). The event can be ignored, just logged or the user can be prevented from subscribing to the application until the license has been updated.
The end result of all of these features is a company full of devices which have no local storage, but have a completely secure PC environment without the headaches of managing the PC environment. Support cases are at worst a reboot of the device which results in bringing the user back to a known state.Application development is much easier because only one environment need be tested and the environment can be easily reproduced when issues with an application arise. Deployment is as simple as creating the application (or application update) on the central server and having that server perform a distribution to all network servers. License monitoring and conformity can be ensured by the administrator using WSM. The overall cost per user can be drastrically reduced, especially if the end clients are replaced by thin clients.
Subscribe to:
Posts (Atom)