REPORT
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
1E.COM
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Contents 3 4
Background Overview: Early Migrators and Their Approach
5
Major Components of Overall Migration Costs and Where to Focus on Speed and Cost Reductions User-centric IT Management and Windows Migration Maturity Model Major Cost Categories for Windows Migrations Application Testing and Certification Hardware Testing and Refresh Application Prep and Delivery for End Users
7
OS Migration Execution Training and Support 9 10 11 13 15 17 18 20 23 24
Share this
1E.COM
In Focus: Application Preparation and Delivery for End Users Manual Mapping Post-OS Migration Install User Driven Self Service of Post-migration Application Delivery Full User Application Mapping In Focus: Migration Execution Manual Brute Force Systems Management/ Manual Hybrid “Zero Touch� Full Systems Management Architecture True Zero Touch: Software Based Distribution, PXE Services and User State Migration Conclusions
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Background Most organizations have been through the challenge of migration from XP to a supported version of Windows (either Windows 7 or 8). For most this was a significant investment in time and resources. 1E helped organizations to minimize this cost and disruption with over 1m PCs migrated to date at a fraction of the cost typical for a large scale migration. Some organization have yet to migrate from XP and are paying the price having taken on the enormously high costs of extending XP support for the near term, and some have even evaluated the risks of moving forward for a period of time in an unsupported state on XP. These late movers find themselves far behind their peers in meeting a hard deadline, and it can appear to them that all of their options are either fraught with extreme cost or extreme risk. And while it may be fair to view them as having reaped the consequences of a failure to act earlier, this trailing group may actually have a benefit they can realize from their position. They have the ability to look at what their peers have done and not done; evaluate those organizations’ successes and challenges; and craft a tight plan that actually puts them in a better cost position than the early movers, with a superior user experience and satisfaction level to boot. With Windows 10 due for release this Summer, organizations that chose to move to Windows 7, as opposed to Windows 8.x, are seriously considering a move to Windows 10 in the short term. However, having just been through the migration from XP, many are concerned about embarking on another costly Windows Migration so soon after the previous one. There is a solution to this... An evaluation of the strategies that organizations used to migrate from Windows XP to date and their associated costs, shows that the very best practice strategies have the triple benefit of being a fastest route to a completed migration, the most user-friendly and likely to delight end users, as well as the most cost effective . Whether you are still to migrate from XP or want to move to Windows 10, 1E's Zero Touch Windows Migration provides the cheapest, easiest migration path and can make your next Windows migration Business-as-Usual.
1E.COM
3
* netmarketshare.com/operating-system-market-share.aspx?qprid=10&qpcustomd=0&qptimeframe=M&qpsp=174) ** Source: Forrester Research Inc. Forrsights Hardware Survey Q3 2013
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Overview: Early Migrators and Their Approach Most organizations that moved early to migrate up from Windows XP had an obvious point of reference in their last major Windows migration – almost exclusively their migration to XP in the first place. For the vast majority of these organizations, the migration to XP was viewed as a bear of a project, one that took considerable time, manpower and cost. Very, very few took a look at their migration to XP and determined to follow the same template. Rather, the common reaction was to set out to conduct a project that was better, cheaper and more efficient by orders of magnitude. One clear challenge to this goal that organizations immediately faced when looking for a marked improvement in their migration experience was that because of the relatively long active life of Windows XP – and resultant lack of need to revisit or rethink migration execution – not much had actually changed in the intervening decade in terms of migration “best practice” processes and methods. While there were some advances in general Operating System deployment (OSD) toolsets and greater reliability and reusability of PC hardware, the most common upgrade methodologies on offer for XP-to-7 migrations were largely the same as for NT migrations.
Somewhat undaunted, many organizations set out to get ahead of the end of life deadline for Windows XP, and started down the path of executing their migrations. At 1E we have had the opportunity to work very closely with a large number of organizations of all sizes on their migrations, including some of the very largest Windows estates in the world. Some have worked with us from the beginning of their deliberations, while others came to us after some of their initial plans failed or got off to alarmingly slow starts. In all cases, we had a unique position to see the planning and cost estimates for a variety of migration scenarios from organizations worldwide, as well as many more after the fact cost calculations after a migration approach was executed for a subset of their users. In this paper, we will survey the landscape of how migrations from Windows XP have been conducted, what the costs were of the various approaches, what methods have provided satisfaction to the end-user community, and draw some highlights around what we have discovered to be the most successful, best practice, highest satisfaction and lowest cost methodologies.
Indeed, Gartner’s estimated per-PC migration costs for XP migrations were virtually unchanged from their equivalent set of numbers for NT migrations (see Gartner publications “Windows 2000 TOC and Migration”* [estimate range $1250-$2050], “Prepare for Your Windows 7 Migration Crunch”** [estimate range $1274-$2069]).
1E.COM
4
* Source: Gartner Research Inc. Getting to Windows 2000 Professional: A Costly Migration by Kevin J. Knox, Michael A. Silver, Michael Gartenberg 15 September 1999 ** Source: Gartner Research Inc. Prepare for Your Windows 7 Migration Crunch by Charles Smulders & Stephen Kleyhans 24 June 2010
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Major Components of Overall Migration Costs and Where to Focus on Speed and Cost Reductions User-centric IT Management and Windows Migration Maturity Model When comparing the costs associated with various potential approaches to a major Windows migration, it is easy to jump directly into the specific buckets of IT costs that make up the whole of a migration project. While it is of course necessary to look at costs at this level of granularity – and we do in great detail in this paper – it is extremely helpful to first consider where these costs fit into the overall service level and user satisfaction targets that all IT organizations strive to maintain. For while those in IT are rightly focused on minimizing the cost of their migration overall, this focus should not obscure an emphasis on value of the migration spend, and therefore the return on that spend in terms of user satisfaction and service levels achieved. What do users want out of IT generally and a Windows migration more specifically? Surely they are not particularly interested in the rationalization of applications, rather they want the best possible application for the job readily at hand so they can dive right to work upon the migration of their user account.
In general, users want three things from IT: • A gility: Users want to be able to achieve maximum productivity without their systems or IT organizations slowing them down or acting as obstacles. • Responsiveness: Users want to know that they are listened to and that their needs remain top of mind for IT. • S elf-service: Users want the independence to act and choose for themselves. They understand that IT must set parameters to manage the security and stability of the organization environment, but once those parameters are set users want and expect to direct their own access to tools and services that help them do their job. In addition to these, power users in the organization take these concepts a step further and typically want some level of ownership of their system, including things like backup and admin rights. With all of this in mind, it is helpful to think of conducting a migration within a user-centric maturity model. The good news represented in this model is that as organizations make choices that are the most cost- and time-efficient, they are also making the choices that meet the needs and expectations of their end users and that ultimately drive high satisfaction and advanced service levels. Let’s now examine how these choices play out in terms of cost, and what kinds of approaches we have seen organizations take to their migration projects to date.
1E.COM
5
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
OS Migration Approaches Maturity Model Maximum Efficiency IT STAFF REQUIRED
Army of contractors
Cohort of contractors
Dedicated systems administrators needed
Business as usual systems administrators
TECHNOLOGY
Men and vans carry image on USB devices
SCCM without OSD functionality
Some degree of automation using SCCM with OSD functionality
Full automation using SCCM with 1E Solutions
SYSTEM MANAGEMENT OVERHEAD
None to minimal
Monitoring and troubleshooting of SCCM infrastructure
Monitoring and troubles-hooting of SCCM infrastructure and SCCM OSD
None
SPEED OF MIGRATION
3-5 PCs migrated per technician per day
3-5 PCs migrated per technician per day
7-11 PCs migrated per technician per day
50-300 PCs migrated per technician per day
FINANCIAL IMPACT
$100 to $275 per migrated PC
$75 to $206 per migrated PC
On average, $85 per migrated PC
On average, $8.5 per migrated PC
Manual brute force
Systems management / Manual hybrid
“Zero-Touch”
True Zero-Touch
USER SATISFACTION
Migration Approaches
1E.COM
6
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Major Cost Categories for Windows Migrations Major enterprise OS migrations are very large and complex projects that can incorporate a variety of cost line items, beyond direct migration activities, in budget documents depending on how an organization allocates costs across different areas of their business on major technology projects. Once those more extraneous items are eliminated however, project costs typically fall into these five major buckets: 1. Application Testing and Certification 2. Hardware Testing and Refresh 3. Application Prep and Delivery for End Users 4. OS Migration Execution 5. Training and Support To briefly define each of these areas: 1. Application Testing and Certification The tasks and activities relating to identifying all applications in use throughout the organization and determining their compatibility for the new OS. This includes, but is not limited to, the selection of applications to bring to new OS and applications to be eliminated from the environment, version rationalization, testing of each intended application and version, and certification process to ensure that applications and OS images do not break on the other side of the migration. 2. Hardware Testing and Refresh The identification, acquisition and testing of all PCs meant to receive the new OS ensures the actual machines are in place for the execution of the migration.
1E.COM
7
3. Application Prep and Delivery for End Users Once rationalized and tested, applications then need to be distributed to the new OS images along with the migration or after the OS migration is completed. This category captures all of the tasks associated with identifying applications for each user or group of users, getting applications ready for distribution and finally distributing the applications themselves. Tasks and processes in this bucket varies widely. 4. OS Migration Execution The actual imaging of each machine, including updating user data, if applicable. 5. Training and Support Captures all end-user communication and preparation and execution of training materials and sessions, as well as standing up and executing a support apparatus for end-user problems and/ or questions.
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Of these, most organizations today – even if they have not yet finalized their plans for how to ultimately execute their migrations – have taken significant steps to accomplish the items in the first two buckets, as they represent foundational prerequisite work that must be squared away and undertaken regardless of how the organization ultimately decides to effectuate the placement of an upgraded OS and applications on end machines. In particular, the second bucket around Hardware Testing and Refresh is just simply not a major differentiator among migration methodologies – in all cases, regardless of how the migration gets accomplished, the proper hardware has to be in place and the ways and means of evaluating the estate and bringing machines up to standard is fairly uniform. Similarly, in the last bucket (Training and Support), there is very little differentiation based on migration methodology. These requirements tend to remain static even as organizations evaluate multiple methods for the on the ground work of migrating machines and applications. The overwhelming majority of organizations who have parallel plans for migration have the exact same tasks, responsibilities, goals and ultimately costs for this area. Therefore, we will not consider this bucket in our migration methodology comparisons.
1E.COM
8
Used Rarely used Unused
13,476
(63%)
896
(4%)
6,880
(33%)
In the first bucket (Application Testing and Certification), in the cases where organizations have not even begun this preparatory work, very real savings can be achieved by conducting optimized rationalization or taking usage inventory up front. Using a tool to determine levels of usage can strip out unnecessary testing and certification activities on applications that – while they may be widely installed – are generally not in use, and where any instances where the application is being used, these applications can be removed and replaced by other applications that offer similar functionality and are more widely used in the organization. Besides this specific scenario where even the most basic prep work on the application side has not been tackled, really the comparison between different migration plans and methodologies comes down to the third and fourth buckets on our list: Application Prep and Delivery for End Users and OS Migration execution. These are the two areas we will focus on in detail for our comparisons of time and cost.
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
In Focus: Application Preparation and Delivery for End Users Even with a perfect migration of a machine and user from old OS to new, it almost goes without saying that a user will only be as productive as the completeness of the migration of his or her applications. Most organizations that have completed migrations from XP have put considerable thought and effort into rationalizing the myriad versions of applications and conducting comprehensive compatibility testing across those versions. Yet over and over again one of the major sources of unexpected effort and cost during a migration project has been the assigning of proper applications and versions to the appropriate users, and ensuring those identified applications and versions are actually delivered to those users and machines.
1E.COM
9
Organizations have tried to handle this challenge in a number of ways, ranging from the purely manual, to leaving it up to the user to fill in gaps, all the way to a fully automated mapping solution. The costs across these different methods vary widely, such that choosing the best and most cost effective approach in this area has become one of the very best ways to cost-optimize an XP migration project for just about every organization.
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Let’s have a closer look now at the most prevalent methods for application preparation and delivery for end users, some of their pros and cons, as well as expected cost per user: Manual Mapping In this approach, a technical resource on the project team conducts an exercise for each user and/ or machine to determine what applications are on the Windows image and therefore what applications should be part of the user’s new OS image. This may be accomplished by the technical resource alone, but more common is a resource sitting down with each user individually to go over their choices to include the user’s input. Taking this step with the user him- or herself is actually a more efficient manual method, as it tends to significantly mitigate the need to take corrective action after the migration if and when a technical resource makes incorrect assumptions and/ or application choices on that user’s behalf. Across organizations that have
conducted this approach, the average effort per user is right around 30 minutes per user. In terms of project cost, using a standard fully-loaded hourly rate of $50 for IT personnel, this translates to a cost of $25 per user or machine migrated. This base cost grows larger when the time spent by the end user is figured into the calculation. Assuming the same rate of $50 per hour, the total cost per user or machine balloons to $50. For a 20,000 PC organization, that translates as a $1 million effort just to accomplish this single step. In addition to the high cost per user, this method is extremely time consuming and can have a significant impact on project and migration timelines.
Technical Effort
Time: Communication with End User (min)
20
Time: Updating Mapping Document (min)
5
Time: Administration and Scheduling (min)
5
End User Effort
Time: Communication with Technical Resource (min)
5
Time: Average for Follow Up Questions (min)
2
Time: Administration and Scheduling (min)
3
Total Effort Time (hours) Fully Loaded Hourly Rate ($) Number of Users Total Cost ($)
1E.COM
20
Time: Preparation for Mapping Discussion (min)
10
1 50 20,000 1,000,000
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Post-OS Migration Install With this approach, the migration project focuses first on completing the migration of each user or machine from XP to the new OS, then handles the delivery and installation of the applications. One of the benefits to this approach is that it can dramatically simplify the OS migration step of the project – indeed, depending on how this general approach is executed it can eliminate the need for tactics like multiple images based on geography, business unit, etc., and provide one single “gold standard” image for the entire organization. However, although arriving at the new OS can be made more straightforward this way, in terms of effort and cost the net effect is to merely kick the work and associated costs down the road. There are typically two execution tactics in carrying out the Post-OS Migration Install approach. The first is to simply not bother with including any applications at all in the standard image. This tactic generally leads to the ability to have just one OS image for the entire organization, but increases the post migration effort. The second identifies a core set of standard applications to include in an image, with post-migration activities limited to filling in the gaps with less widely distributed or more specialized apps following the migration. In either case, organizations who have attempted this approach report that users experience extremely poor user satisfaction. Between not hitting the ground running on their work immediately after migration, as well as downtime associated with the installation of the applications, users are much more likely to view their migration experience as a major inconvenience using these methods.
1E.COM
11
A seeming major benefit to this approach – that the time to complete the OS migration part of the project can be accelerated to meet XP end of support deadlines – actually is illusory. Organizations cannot simultaneously move mass numbers of users and machines to the new OS while accommodating users’ needs on the application side. In addition, there would be significant productivity hits to the organization by moving large numbers of machines all at once without all of the business critical applications installed on end user machines. The complete breakdown of mission critical business processes would be all but assured in this scenario. In terms of cost impact of this approach, those organizations who have conducted it have typically found that with the “no application in image” method, there is a technical resource effort of about 20 minutes per PC to get all applications back out to users, plus end user downtime of anywhere from 30 to 90 minutes in the weeks and months following the migration all told. Splitting the difference on this range and using our standard hourly rate of $50, this results in a cost of $66.67 per machine migrated, or a hit of $1,333,333 to our 20,000 PC example organization. For the “standard applications in image” method, costs are a little lower. The technical resource effort is shaved by five minutes (the lion’s share of the effort is not around delivery of standard apps, rather the more specialized apps take the most time to track down, identify and deliver), and the user downtime drops to a 40 minute average. So the hit in this case totals out to $916,667, or $45.83 per PC migrated.
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
No Applications in Image Method Technical Effort
Time: Time to Review User Needs (min; per user) Time: Average Time to Distribute Application (min; per app)
10 .5
Average Number of Apps (per PC)
20
Time: Total to Distribute Applications (min; per user)
20
End User Time Lost
Time: Average Downtime (min; per user)
Total Time Spent (hours) Fully Loaded Hourly Rate ($) Number of Users Total Cost ($)
60
1.33 50 20,000 1,333,333.33
Standard Applications in Image Method Technical Effort
Time: Time to Review User Needs (min; per user) Time: Average Time to Distribute Application (min; per app)
10 .5
Average Number of Apps (per PC)
20
Number of Standard Apps Pre-Installed
10
Time: Total to Distribute Applications (min; per user)
15
End User Time Lost
Time: Average Downtime (min; per user)
40
Total Time Spent (hours)
.92
Fully Loaded Hourly Rate ($) Number of Users Total Cost ($)
1E.COM
12
50 20,000 916,666.67
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
User Driven Self Service of PostMigration Application Delivery This approach is an offshoot of the Post-OS Migration Install method and can provide some significant benefits, though there may be productivity hits both up front immediately following the OS migration as users struggle to get up and running, as well as down the line when application needs pop up that were not present right at migration time. Here, instead of technical resources doing the work to deliver applications to end users, the users are directed to a selfservice portal where they select and kickoff the installation of their needed applications. This approach brings down the technical effort significantly, and would also reduce the user downtime for either application/ image scenario by about 50%. Ultimately for the “no applications in image” scenario, the total cost per PC migrated calculates to $25.40, while for the “standard application in image” scenario, the per PC cost drops to $17.07.
1E.COM
13
For a 20,000 PC organization, this represents a savings on this single step of $500,000 to $700,000 over the standard Manual Mapping approach. These costs are exclusive of the acquisition or development of a Self-Service tool, but do include time spent by the project team deploying and configuring the Self-Service tool.
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Self-Service: No Applications in Image Method Technical Effort
Time: Total to Deploy and Configure Self-Service Tool (hours)
160
End User Time Lost
Time: Average Downtime (min; per user)
30
Total Time Spent (hours)
.51
Fully Loaded Hourly Rate ($)
50
Number of Users
20,000
Total Cost ($)
508,000
Self-Service: Standard Applications in Image Method Technical Effort
Time: Total to Deploy and Configure Self-Service Tool (hours)
160
End User Time Lost
Time: Average Downtime (min; per user)
20
Total Time Spent (hours)
.34
Fully Loaded Hourly Rate ($) Number of Users Total Cost ($)
1E.COM
14
50 20,000 341,333.33
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Full User Application Mapping In this approach, a fully automated solution is deployed that reads and identifies the applications and versions installed on the Windows image and maps each of those to a selected application and version on the new OS, based on rulesets created by the organization. For example, an organization could see a version of an application in XP that is not compatible with the new OS, and based on the user’s or machine’s business unit, map it to the new version of that application, or to a freeware alternative, or to no application install, depending on the business need. Using the Application Mapping capability available in the integration of the 1E Shopping and AppClarity
So if User A heavily used a non-Windows 7 compatible version of Adobe Acrobat, but User B hardly used that same version at all, User A would automatically map to a new compatible version, while User B would get a freeware alternative. This approach really addresses the three key goals: cost reduction, positive user experience and migration project speed. It barely requires any intervention at all from the end user, so that the end user feels like they have had to invest very little of their own time. All migrations are completed with all needed applications fully installed, eliminating extra steps, costly effort and user downtime.
solutions, organizations could further create rule sets based on the usage profile of the application on the user’s Windows machine.
LEGACY OS BUILD
1E ANALYTICS
Adobe Photoshop CS5 AppClarity & Shopping
1E.COM
15
NEW WIN 10 BUILD Used
Adobe Photoshop CS6 ($500)
Rarely used
Paint.NET (Free)
Unused
Do not install
=
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
From a cost perspective, the per PC cost drops to $1.20 per PC, and represents a savings of $976,000 over the Manual Mapping method, again exclusive of costs to develop or acquire the toolset, but including all project team time to deploy and configure the toolset for automated application mapping.
Full Application Mapping Technical Effort
Time: Total to Deploy and Configure Self-Service Tool (hours)
160
Time: Total to Deploy and Configure App Inventory Tool (hours)
100
Time: Total to Configure App Mapping Integration (hours)
220
End User Time Lost
Time: Average Downtime (min; per user)
Total Time Spent (hours) Fully Loaded Hourly Rate ($)
1E.COM
0
.02 50
Number of Users
20,000
Total Cost ($)
24,000
16
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
In Focus: Migration Execution After all of the preparatory steps and the work done in support of a migration project, the actual process of physically migrating a user or machine to a new Windows OS is quite obviously the heart of any migration effort. At the end of the day, organizations must schedule and then execute every user and machine targeted for migration. Often, organizations have processes in place for general Operating System Deployment (OSD) that they look to leverage in a migration project. A very large number of these organizations come to learn that processes meant to accomplish day-today OSD in support of new machine issue or break-fix scenarios don’t measure up to the rigors and demands presented by a mass-scale enterprise OS migration. Compelled back to the drawing board, organizations have come up with a stunningly wide array of methodologies, some of which reflect a desire to optimize the process, while yet others seek to avoid complexity and the introduction of risk.
1E.COM
17
In almost all cases, internal business review of these projects result in severe sticker shock, whether at the dollars involved, the number of required boots on the ground, the planned duration of the project, or some combination of these three considerations. Organizations with whom we’ve reviewed their migration plans have submitted literally dozens of potential approaches and plans, but by far the most prevalent approaches across all industries are: 1. Manual Brute Force 2. Systems Management/ Manual Hybrid 3. “Zero Touch” Full Systems Management Architecture 4. True Zero Touch: Software Based Distribution, PXE Services and USM Let’s look at each approach in some detail.
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Manual Brute Force This is probably the most straightforward approach in terms of description. Indeed, this approach is what is typically proposed by many services firms, largely because it is something that they provided for the customer during a previous migration, and – even if costly and long – did at the end of the day get their estate migrated.
The costs here tend to depend on two major factors: 1. The ability of the geographic distribution of hired contractors to be made very close to the geographic distribution of the PCs to be migrated and; 2. The rate of machines each contractor
This approach means connecting a technical resource with a machine, either by deploying the resource to the machine site or shipping the machine to the technical resource, and having the resource reimage the machine using media – usually a USB device. In order to do this in 12 months or less, most large organizations have to hire an army of contractors to carry out this plan. Organizations who have planned or executed a migration method like this have forecasted or recorded costs ranging from $100 to $275 per migrated PC. Using our 20,000 PC organization as an example, the range here is $2 million to $5.5 million.
can migrate per working day. For the former, if it is not feasible to match the geographic distribution, travel and/ or shipping costs become a major cost driver. For the latter, we tend to see somewhere on the order of three to five machines migrated per technician per day. Costs on the high end of that scale are therefore significantly smaller than on the low end. Referencing back to our Maturity Model above, this really represents a nearly primitive migration method. Conducting a brute force migration essentially means a deskside visit or shipping of a large portion of machines, or both, and all of these scenarios are incredibly costly. In addition, this method almost always ensures a very long project duration. So why do organizations go down this route, even fully aware of these limitations? In looking to minimize risk, they simply pick the safest option, not realizing that the reliability of newer, faster, more mature methods are actually more reliable.
1E.COM
18
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Please note: Costs are typically 25% lower than shown for Systems Management/ Manual Hybrid method, depending on number of PCs subject to each side of the hybrid approach.
Top End
Technical Effort
Time: Hardware Prep and Connectivity Check (hours)
.25
.5
Time: User State Data Backup (hours)
.5
1
Time: Imaging (hours)
.5
1
Time: State Data Restore (hours)
.5
1
.25
.5
Time: Post Imaging Check (hours) Total Time (hours) Fully Loaded Hourly Rate ($) Time Cost ($; per PC) Additional Spend: Shipping ($)
2
4
50
50
100
200
–
75
Total Cost ($; per PC)
100
275
Total Number of PCs
20,000
20,000
2,000,000
5,500,000
Total Cost ($)
1E.COM
Low End
19
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Systems Management/ Manual Hybrid Here, organizations seek to augment their systems management and “everyday” OSD capabilities with a smaller contractor army to make up the gap. Typically, these organizations target a certain subset of the environment (usually the main company HQ) to migrate via the existing systems management OSD tools, while using manual brute force to effectuate the migrations of the rest (usually the farther flung branches and mobile employees). This approach tends to fail to mitigate full manual brute force method costs for three main reasons. First, the existing OSD processes and architecture inevitably have to be enhanced in some way in order to see any advantage in level of effort to migrate a machine or project duration, so the attempt to split the difference forces the organization to get the worst of both worlds. Second, very often the machine groups that would benefit the most from a centralized process are the farther flung machines that end up subject to the Brute Force Method. Third, the gains organizations typically see from native systems management OSD capabilities – even when enhanced – are not the order of magnitude difference they may have been expecting.
1E.COM
20
As detailed in the section on “Zero Touch” Full Systems Management Architecture below, our experience with migrating organizations is that the three to five per technician per day number only marginally improves using native systems management OSD even at full throttle. Cost savings against Manual Brute Force are about 25% per PC migrated, such that our 20,000 PC organization would still be spending $1.5 million to $4 million to execute the physical migration.
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
“Zero Touch” Full Systems Management Architecture This approach has organizations use their prevailing systems management platform as a foundation to build an enterprise scale mass OSD architecture that allows for remote booting and migration of user state data, all centrally managed and executed. It is the recommended migration approach of nearly every major systems management platform, especially including the most widely used platform – Microsoft System Center Configuration Manager (SCCM). In addition to the usual SCCM OSD and content distribution infrastructure – largely made up of distribution points – the migration architecture also requires the deployment of additional architecture for remote booting (PXE service points) and user state data (state migration points). The benefit is widely understood to be that a centrally managed migration can be effectuated once this architecture is in place, which allows for a “zero touch” experience where scheduled machines are migrated remotely from a central location. In practice, however, there are significant realities and costs that many, many organizations learned about only when well into their migration planning (and in some cases, execution). First is that in addition to standing up these various pieces of systems management and OSD architecture, these machines must be monitored throughout the duration of the physical migration of machines. Should they fail (which happens with some regularity), they also need to be troubleshot and repaired. Very often, organizations do not plan for this piece of effort before they plan to go the full systems management route.
1E.COM
21
Secondly, across the many major organizations we worked with in designing, advising and executing their XP-to-Windows 7 migrations, the average number of machines migrated per technical resource per day has averaged about seven per day, with none reporting results better than 11.2 per day. So while the pace of migration using an identical crew is improved using this method, it is hardly a sea change. Coupled with the additional cost of setting up, monitoring, troubleshooting and repairing the extensive OSD architecture, this method has historically sat right about at the low end of the typical scale for the Systems Management/ Manual Hybrid, and moves quickly up that scale if the organization is widely distributed (and therefore has a need for more extensive architecture). So while the migration occurs in a centrally managed fashion and should usually take a marginally shorter time, it does not deliver significant cost savings. As detailed in the tables in section VI, our 20,000 PC organization, equally distributed between large and small locations and with an average ratio of PCs to distribution points, shows a cost of $1.27 million using this method.
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Total Users to Migrate
Core – Wired
Branches
Total
10,000
10,000
20,000
OSD Setup/Config (hours; per DP)
4
PXE Setup/Config (days; per Core Site)
.5
PXE Setup/Config (days; per Branch)
.5
Setup/Config (hours; per SMP) Servers (DP, PXE, SMP) requiring monitoring on given
8 10
migration day (%) Average time for DP, PXE, SMP monitoring (mins; per server) Major troubleshooting (instances; per month) Time to resolve major troubleshooting issue
30 2 16
(hours + travel time)
Number of Core Sites
10
Number of Branches
50
Number of DPs Number of Users per SMP in Core
60 1000
Hourly Rate for SCCM Architect ($)
75
Hourly Rate for Deployment Tech ($)
50
Projected Project Length (months)
12
Projected Lead Time for Architecture/Testing/etc (months)
3
Machines Initially Migrated in Lead Phase (%)
5
Main Migration Period for Remaining Machines
9
(months – calculated)
Migration Architecture Costs Without Nomad ($) DP Setup and Configuration for OSD
18,000
Setup and Config of PXE Environment for each site
18,000
Setup and Config of State Migration Point for each site
36,000
Ongoing monitoring of DPs, SMPs, and PXE during migration Troubleshooting (saturated links, maxed SMPs, etc)
Total Cost
162,000 28,800
262,800
Anticipated FTEs for Baseline Project
14
Projected Project Length (months)
12
Projected Lead Time for Architecture/Testing/etc (months)
3
Machines Initially Migrated in Lead Phase (%)
5
Main Migration Period for Remaining Machines
9
(months – calculated)
Total Users to Migrate Number Migrated in Lead Phase Remain to Migrate Main Migration Phase (months)
Needed to Migrate (per day) Daily Migrations (per FTE)
Total Migration Tech Cost ($) Total Architecture Cost ($) Total Cost: Full System Management Architecture Mehtod ($)
1E.COM
22
10,000
10,000
500
500
9,500
9,500
9
9
53
53
106
3.785714286
3.785714
7.571429
1,008,000 262,800 1,270,800
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
True Zero Touch: Software Based Distribution, PXE Services and User State Migration With a software based solution, all of the wide content distribution, PXE services and user state migration capability that are available from the full systems management architecture approach is made available with a single set of software tools. The effect of this is that it takes away the need to deploy, monitor, troubleshoot and repair dozens or hundreds of different pieces of architecture, and reduces all of these activities to a single instance to work with. This dramatically reduces the effort and cost in setting up the zero touch capability, and extends peak performance to locations that might not even be suitable to set up physical architecture in the first place.
Daily Migrations per FTE with Nomad Daily Migrations with Nomad Total Migration Time (months) Total Migration Tech Cost – Nomad ($) Nomad Deploy and Monitoring Nomad Throughout MIgration ($)
Total Cost ($)
1E.COM
23
In addition, the software solutions tend to be much more reliable than any piece of DP, PXE or SMP server, so the need to troubleshoot and repair is essentially eliminated. This reliability extends further to a much more highly scaled and rapid rate of deployment and migration. At 1E, those using our Nomad solution set have seen rates of migration as high as 200 PCs per technical resource per day, which is almost 20 times the very best using the Full Systems Architecture approach have been able to achieve. If we dial down that expectation to be highly conservative, and instead project 25% of that top end number, we see 50 migrations per technical resource per day, and a total cost of $170,000 for our 20,000 PC organization, not including the cost of the Nomad solution, but including internal technical time to deploy and monitor the solution throughout the migration activity.
50 700 1.36 152,000 18,000
170,000
THE AUTOMATED MIGRATION: AN ANALYSIS OF OPTIONS
Conclusions While the majority of organizations have already made the move to migrate away from XP, as noted at the outset a surprising number of organizations are still trying to figure out their plan. Fortunately, these trailing firms can learn much from both the organizations that have gone before them as well as solution and services providers like 1E who have had the unique opportunity to work with dozens of these organizations at every point along the migration road. What these data points show is that there are very real savings opportunities for organizations who focus on programmatic automation of both the physical migration of machines via advanced content distribution solutions attached to leading systems management platforms and the mapping of applications in the new OS to each user or PC based on applications present in the old OS. With a new migration to Windows 10, following these best practices can save organizations millions and there are essentially zero barriers to using these methods, as providers such as 1E have solutions and expertise at the ready.
About 1E 1E is the pioneer and global leader in Efficient IT solutions. 1E’s mission is to identify unused IT, help remove it and optimize everything else. 1E Efficient IT solutions help reduce servers, network bandwidth constraints, software licenses and energy consumption.
Contact us US: +1 866 592 4214 UK: +44 20 8326 3880 India: +91 120 402 4000 info@1e.com
Share this Š Copyright 2015 1E. All rights reserved. The information contained herein is subject to change without notice. 1E shall not be liable for technical or editorial errors or omissions contained herein.
1E.COM
1E, the 1E logo, NightWatchman, Nomad Branch and Drowsy Server are registered trademarks of 1E. The names of actual companies and products mentioned herein may be the trademarks or registered trademarks of their respective owners.