Column 2

Appian Analyst Briefing: 2009 Overview and Future Outlook

Appian issued a press release last week on their growth in 2009, and had an analyst call today to provide more detail and answer questions. I attended their user conference in October, and was interested to hear their plans in the wake of recent BPM acquisitions.

In short, their 2009 performance was the best in their history:

67% increase in software license revenues
59% increase in international revenues, expanding beyond their UK base to Australia, New Zealand and the Middle East (which shows an obvious bias towards English-speaking countries that they’ll need to better address at some point)
112% increase in number of new customers, including a significant win at Amazon over Lombardi and Pega
Signed 6 major VAR/OEM relationships for both on-premise and SaaS products, including RICOH’s business process automation group

In addition to new product releases and their cloud-based offering, Appian Anywhere (which is now responsible for 10% of their revenue), they’ve productized their professional services framework and implementation methodology, and have launched a free (but closed) online community for their customers and partners. Although they still have a significant base in the US federal government, deployed in 22 agencies and departments, they’ve expanded into financial services, insurance, telecommunications and logistics.

Matt Calkins gave us his view of their future, starting with “Appian Is Not For Sale” and contrasting their position of almost 100% self-funded growth, with only recent venture capital infusions that will not force their hand any time soon, against that of recently-acquired BPM vendors who may have run out of road with their long-time VCs and forced to sell. He sees the BPM fight as now being between themselves and Pega, pitting their rate of innovation and ease of use against Pega’s dominant market share. The stack vendors are certainly serious competition, but a customer’s decision to go with a stack vendor versus a BPM suites vendor is usually made so early in the evaluation cycle that Appian rarely finds themselves in a short list head-to-head against a stack vendor. I found that to be a refreshingly realistic view of the market: BPM isn’t a homogeneous market where every customer always looks at every vendor; the vendors are passed through many filters along the way, and the true battles are between those that end up on the same short list. The higher-level strategy, of course, is to change those filters.

Appian will be a company to watch this year, as one of only a few remaining players in a still very competitive BPM space. They would be well-served by opening up their online community to non-customers (although possibly reserving product-specific portions of it for customers) in order to better show off their market leadership. They’re also in a position to achieve dominance in the SaaS BPM market (which they already claim to lead), although there’s still a lot of discussion about the actual utility of cloud-based BPM.

IBM-Lombardi Deal Closes

That was fast! Proving my predictions to be ever so inaccurate when I said that it could be months before this closed, the IBM acquisition of Lombardi has closed less than six weeks after it was announced. Good news for IBM, Lombardi, and Lombardi customers.

Not surprisingly, IBM has toned down the “departmental” rhetoric that they initially used when talking about how Lombardi fits into their portfolio; there’s now only a mention of departmental managers rather than branding the software suite as being targeted only at departmental applications:

Lombardi adds a new dimension to IBM’s enterprise-wide BPM capabilities by giving organizations the ability to quickly adjust their business processes to support sudden and changing needs, especially those that rely heavily on collaboration to complete a task or project. Lombardi strengthens IBM’s capabilities in automating these processes, while empowering managers at the department level to change already running processes on the fly, eliminating the need for complicated and time-consuming technical intervention.

Of course, this now makes BP3’s upcoming bpmCamp an IBM user group conference. 🙂

CrisisCampTO Planning Meeting

A bit off topic for my usual blogging here, but I spent this afternoon at the initial planning meeting of CrisisCampTO, the Toronto manifestation of Crisis Commons. Although this is happening here and now in response to the earthquake disaster in Haiti 12 days ago, Crisis Commons has a broader mandate:

We are an international volunteer network of professionals drawn together by a call to service. We create technological tools and resources for responders to use in mitigating disasters and crises around the world

We’re here today to work on anything that can be done to help, in collaboration with other Crisis Commons teams all over the world, on the various projects that have been defined by Crisis Commons based on requests from NGOs to fill a need that they have. The bulk of the projects fall under the category of software development, but there are also teams for social media, logistics and more general duties.

Our first goal today is to find a development project for the bulk of the Toronto team to get involved with, and learn how to plug into other Crisis Commons groups around the world. There is quite a bit of infrastructure already in place to connect up, including IRC channels (retro, I will definitely need a refresher course) and voice conference lines, plus a rapidly growing wiki.

I have a pretty broad range of skills to apply here: although I don’t really write code any more – unless I’m really inspired – I can do all the other stuff around development (requirements, testing, documentation). I also do a lot of social media stuff, and have attended more unconferences than you can shake a stick at, so can help with the local social media efforts such as wiki gardening, Facebook and Twitter updates, and more.

The main goal of today is to get ready for next Saturday’s CrisisCampTO (time and venue to be announced shortly), by getting some basic team structure in place and selecting one or more projects to which we will be contributing. That way, when newbies show up next week, they can start contributing immediately.

One of the things that we learned about today is Sahana, an open source disaster management system that was created in response to the Sri Lanka tsunami in 2004. There’s a Sahana instance set up just for Haiti, although it still needs a lot of content added, and possibly some development to add specific requested functionality. We also saw OpenMRS, an open source medical records system, and Ushahidi, an SMS-to-web service that accepts requests for assistance sent by text message to a specific shortcode, and makes them available to aid agencies. If you check the feed from Haiti, you can see requests for food, water and medical assistance that have been received, translated if required, and logged for followup. In summary, there are a ton of free, open source projects that can be applied to the Haiti disaster; some of them as is, others requiring some customization. This is were we all come in.

More BPM Acquisitions: Progress Buys Savvion

BPM acquisitions must be in the air: today, Progress Software announced that they’ve bought Savvion for $49M. This is hot on the heels of IBM’s announcement last month that they’re buying Lombardi, with one huge difference being that Progress doesn’t already have a BPM product in their lineup, whereas IBM has two. Of the three mid-range BPMS-only vendors that I would most commonly name – Appian, Lombardi and Savvion – that’s two out of the three announcing acquisition in less than a month. With the economy just starting to pull out of a huge pit, that’s telling news: as I mentioned in my post about Lombardi, if the economic climate were different, these would be IPOs that we’d be seeing rather than acquisitions. These acquisitions by larger companies, however, changes the BPM market landscape pretty significantly, since this makes it significantly easier for Lombardi and Savvion (under the IBM and Progress banners, respectively) to get a foot in the door of larger customers who rely on their major vendors to bring them enterprise solutions, rather than considering a smaller company. One advantage that Progress/Savvion have at this point in time is that the acquisition is actually closing today (or later this week), whereas IBM/Lombardi went the pre-acquisition announcement route, and will endure several months of limbo before the deal closes. [Update: I’ve received a few tweets and emails indicating that the IBM/Lombardi close will happen very soon, possibly around February 1st, although I haven’t heard a final date. My “several months” was based on past experience.]

I had an early morning call with Dr. John Bates (CTO of Progress) and Dr. Ketabchi (CEO of Savvion), but a few people obviously had earlier time slots: Neil Ward-Dutton has already posted his initial thoughts, as has Jason Stamper. I agree with Neil that this is a smart move for Progress: a good fit of products with minimal overlap, directly addressing some of the challenges that they’re hearing from their customers in terms of achieving operational responsiveness. The existing suite of Progress products allows for determining what happened within an organization – a rear-view mirror approach – but not much that allows the organization to quickly change how they’re doing things in order to drive efficiency or respond to changing conditions. Bringing BPM into the fold allows them to change that, primarily through tying Progress’ Apama CEP with Savvion BPM, but also by leveraging the rest of the Progress SOA and ESB infrastructure, including data and application integration.

Savvion’s had a couple of internal shakeups in the past two years: in early 2008, Savvion axed contractors, most of their marketing department and some salespeople, ostensibly in order to shift towards a solution focus, although at the time I said that they could be positioning themselves for acquisition. They’ve had a strong push on their vertical solutions since that time, wherein they develop frameworks for vertical applications, then allow partners – or even customers – to built vertical solutions on those common frameworks.

Like many BPM vendors, Savvion has often sold to the technology side of organizations but have shifted focus to the business side recently. Progress is still a very technology-focused set of tools, so it will be interesting to see how well they can bring together the different marketing messages. In my conversation with him this morning, John Bates said that they’re moving towards more of a solutions-oriented approach rather than product-oriented: although this is an easier sell to the business side, it can be used to mask a number of disparate products being clumped together without much natural cohesion (cf. “IBM BPM”).

There will need to be some product integration points to be able to really sell this as an integrated suite of tools rather than a “solution” patched together with professional services. First, they need to bring together a common process modeling environment. Ditto for an event/process monitoring environment. Third, they need to consider the touchpoints within application development: although data integration and application integration will be designed using the existing Progress products, these have to be seamlessly integrated into Savvion’s process application development environment. There are likely also areas of integration at the engine level, too, but getting the developer and analyst-facing tools integrated first is key to acceptance, and therefore sales, of an integrated solution.

Another consideration will be a software-as-a-service offering: Savvion already has inroads in this with their BPO market, although they haven’t yet announced any consumer-facing SaaS products. Bates stated that Progress considers SaaS “an important paradigm”, which I would translate as “we know that we have to do it, but aren’t there yet”. Pushing BPM and CEP to mid-range and smaller companies is going to require a strong SaaS offering, as well as providing a platform for larger enterprises to use for piloting and testing.

Because the acquisition has already closed, or is closing within the next day or two, Progress and Savvion sales and partner channels are already being brought together; the same will happen soon for marketing teams. As always happens in this case, there will be some losses, but given the small degree of overlap in product functionality, they’ll probably need most of the skills from both sides to make this work. Dr. K. has stated that he’ll stay with Progress, although his role hasn’t been announced.

The BPM+CEP equation is becoming increasingly important as organizations focus on operational responsiveness, and I think that it’s particularly significant that Progress appointed Bates – formerly co-founder and CTO of Apama before their acquisition by Progress – to the CTO position during the time when they must have been negotiating to acquire Savvion. Clearly, Progress sees BPM+CEP as an important mix, too.

Disclosure: Savvion has been my client within the past year, for creating a webinar and internal strategy reports, although we have no active projects at this time.

Participate in a BPM Research Survey

Strangely, this message was sent to the BPM Arena Google Group on December 7th, but just delivered to my email this morning. Two students from the University of Applied Sciences
Ravensburg – Weingarten are doing a research survey about BPM usage, and are looking for participants.

I realize that many people have left for the holidays already, but if you haven’t, you probably have plenty of time to take the 15 minutes required to fill it out. Spread the Christmas cheer, and help a couple of students with their research:

We are 2 Bachelor students from the University of Applied Sciences
Ravensburg – Weingarten in Germany.

In cooperation with the Technical University of Eindhoven
(Netherlands), we are performing a study, in which we analyze possible
correlations between organizational characteristics and the success of
BPM projects.

In order to analyze selected issues, we decided to conduct an online
survey.

We write you today to ask you to participate in this survey if you
have experiences in using BPM technologies or if you have
participated in BPM projects. Almost all questions are simple
multiple choice questions.

Their answering should not last more than 15 minutes. The
questionnaire is available via the following link:
http://sites.google.com/site/bpmsurveyen/home

The survey is online until Christmas 2009. The evaluation of the
survey results is anonymous.

Each participant that denounces a valid email address will get a free
report that summarizes major results of the survey. This report will
be sent to the participants in March 2010.

If you have questions regarding the study, feel free to contact us by
email to:
[email protected] and [email protected]

We would be very grateful if you would help us with our study.

Thanks in advance for your support and kind regards.

Edgar Rot and Christian Waibel

As they mention in their email, if you include your email address when filling out the survey, you’ll receive a summary of the survey results.

IBM Buying Lombardi: A Bauble on their BPM Christmas Tree?

I was on the analyst call this morning to hear about IBM’s acquisition of Lombardi – a pretty significant acquisition in the BPM space. Lombardi is the best known of the mid-range BPMS vendors, and if the economic climate weren’t quite so dreary, I imagine that they’d be doing an IPO rather than being acquired, or at least staying as an independent rather than become part of an organization that offers what Phil Gilbert (president of Lombardi) recently described as not BPM, but “Orwellian marketing rhetoric”. Given that Phil has done everything except call IBM the “evil empire”, it’s hard to imagine the drivers behind this acquisition.

Lombardi will become part of the WebSphere team, which is what I said that they should have done with FileNet’s BPM three years ago when that acquisition happened; maybe they’ll do this one right so that the product actually becomes an integral part of IBM’s BPM offering rather than a poor cousin. Obviously, the motivations for this acquisition are much different than that of FileNet: in FileNet’s case, they were buying it more for the content management capabilities, and I think that they just didn’t know what to do with the BPM product except reposition it as content-centric and leave it there; they were also buying a huge portfolio of FileNet enterprise customers where they were directly competing. With Lombardi, however, they’re clearly after the great software, both Blueprint for collaborative process modeling (which will benefit GBS, IBM’s professional services arm) and Teamworks for human-centric process execution. IBM already has a strong integration-centric offering with WebSphere Process Server (WPS), to which they’ve been attempting to add human-centric capabilities; done right, they will slide Teamworks in as a fully-integrated, human-centric part of WPS rather than keep it as a separate product offering. Neil Ward-Dutton’s first reactions this morning (prior to the analyst call) make some excellent points about the overlap in IBM’s BPM portfolio:

Although the strengths of Lombardi’s tools are different from IBM’s there is almost 100% product overlap. What’s more the design philosophy of Lombardi’s offering is almost diametrically opposed to that of IBM’s offering – many of Lombardi’s strengths come from its tightly-integrated toolset and repository. It’s not straightforward to see how these things can come together to form a coherent portfolio – unless they’re basically fenced off from each other and positioned as supporting different kinds of BPM scenario (with Lombardi focused on “people centric” processes, WebSphere on “system centric” processes). That brings its own challenges though, particularly for FileNet.

When Judith Hurwitz asked about the alignment of this with the existing IBM BPM portfolio, IBM responded with how they’ve done a lot of work already to integrate Lombardi with WebSphere (which, of course, doesn’t answer the question), then firmly upheld IBM’s position of having multiple overlapping products that will make it difficult for customers to decide which product is best for them, and easy for customers with a big pocketbook to buy multiple products to do almost the same thing, much as they do now with FileNet BPM and WebSphere Process Server. Sure, there are some use cases that are particularly well-matched to one of the three, but Lombardi’s position has always been (and many of the analysts agree, to a point) that they can manage the full range of BPMS functionality, and adding the ability to use the WPS underpinnings for better SOA management governance should be used to improve that position, not make it a competitive offering.

Amy Wohl asked about how IBM will position the Lombardi offering within the market; they see it as a BPM entry route into departments, which might then be shifted to another BPM platform as it works out. IBM’s GBS sees it as filling the layer between business and application development, rather than as an end in itself, and one of the IBM slides stated that Lombardi “brings simple, collaborative and business-driven capabilities for departmental BPM”. That seems to indicate that IBM plans to use this as a relatively low-end offering to get their foot in the door, then bring in the big WPS guns along with a huge portfolio of GBS services. That seems completely counter to the Lombardi philosophy, although I’m sure that they’re not driving the bus at this point.

Tony Baer asked about Blueprint versus BlueWorks; although there is some overlap, I see this as less of an issue since BlueWorks has been somewhat light on process modeling and more about the community. To do this right, I think that they can bring Blueprint’s modeling capabilities into BlueWorks community to replace the BlueWorks modeling; if not, there’s going to be a lot of confusion around this.

James Governer asked about the potential for integration with Lotus, where IBM’s collaboration capabilities are centered; it seems that Lombardi is being positioned as part of the WebSphere suite, so although there might be some synergies (as IBM replied in a somewhat hand-waving way), I’m not sure that that potential will be fully realized.

You can find IBM’s press release here, Lombardi’s press release here, and Phil Gilbert’s blog post about the acquisition here which, somewhat shockingly, finishes with “And remember, like I’ve always said, ‘if it’s BPM, it’s IBM.’”, an interesting counterpoint to his post that I refer to above where he states the “IBM doesn’t do BPM” as part of his long-running anti-stack vendor campaign. That about-face is definitely Orwellian.

I did have to laugh at the email from IBM to analysts this morning, which invited us to tweet using the hashtag #BPM, as if this were the only thing happening in the BPM space today. On the other hand, maybe it is.

OutSystems Agile Platform 5.0 Adds Process Modeling

I’ve reviewed a lot of BPM systems, and one of the most common weak points is in developing the part of the process application that isn’t generated from the process model: user interfaces, data models, security, business logic and integration with legacy systems. Some of this is improving, with the addition of graphical forms builders for UI, and integration of business rules engines, but these are typically the places where you have to drop out of that lovely BPMS environment and write some Java UI code, or hand-craft a data model. Although the process models might be easy to change, as soon as you start writing code, your BPMS turns from agile to brittle.

BPMS vendors tend to focus on process modeling and generating an executable process from that model first, then later tack on features to expand into a full application development suite. OutSystems has taken the opposite approach: they’re an Agile application development suite vendor that’s adding business process management as a feature in their application development toolkit.

Backing up a bit, OutSystems makes a model-driven application development suite called Agile Platform, that allows for rapid development of web-based applications that involve complex system integration and rich user interfaces. Headquartered in Portugal, they developed a strong European customer base before entering the North American market in 2007, and helped boost adoption by offering a free community edition. Agile Platform manages the whole application lifecycle: following the model-driven development, you deploy and manage your applications directly from it, and there are tools for collecting user feedback directly within the executing applications to feedback to the developers for potential changes.

With Agile Platform 5.0, released earlier this week, they’ve added business processes to the application development mix. Since their product is targeted at developers, their process modeler isn’t for a non-technical business analyst: it’s a capability within the model-driven application development environment, just like data modeling and UI development. Other application artifacts, such as screens, can be integrated with the process to show the the interrelationship between the process and application.

The process modeler isn’t BPMN, but a simple flowchart format without swimlanes, and with a small but adequate set of flow elements: start, conditional start, human activity, automated activity, execute process, send email, wait, decision, end and comment. The start event may be triggered by user input on a form or by some simple events include database triggers, but there are no intermediate or end events. Since you’re working in an integrated application development environment, there are data modeling tools allowing you to create the process instance data model and view it as an entity-relationship diagram: this is just part of the overall application data model, and treated as such. Similarly, you can create a user interface, drag fields onto it from the data model, add the screen flow logic, and bind it to a step in the process. If you want a quick demo, there’s a screencam on their site showing how to build a business process in 30 minutes (although it only runs 9 minutes since there’s some fast-forwarding through the boring bits).

The process and application are deployed as a single unit for execution, since the process is just an extension of the application. Agile Platform generates and compiles standard .NET or Java code from the models, updates databases with the data models and deploys the code on the application servers in a single step. If you want to stop using Agile Platform at some point, you can take the generated source code (which looks pretty clean) and work directly with it. Since this executes as compiled code, it’s fairly efficient: they reported one of their beta customers as having 16 high-level processes with more than 200 activities each, and 3.8 million activities in flight at any given time.

An interesting user interface feature is that if you’re anywhere in the application/portal and you have uncompleted tasks, you’ll see a little floating box showing a counter of your incomplete tasks; click on it to list the tasks, then click on one to open and process it. After completion, you’re automatically returned to where you were in the portal before processing the task. That means that occasional process participants don’t need to keep checking their process inbox to see if they have tasks assigned to them.

They’ve also included a predefined analytical model for process reporting and have some monitoring capabilities in the enterprise edition, although I didn’t see that during the briefing.

To be clear, Agile Platform does not offer full BPMS functionality: there’s no true orchestration, no event handling (aside from a few start events), and no process reusability. Their intention is to provide “good enough” process management for primarily human-facing processes, in the context of a web-based business application. The application is primary; process is just another feature of the application. Although process-centric BPM types might recoil in horror from that idea since it violates the idea of end-to-end processes independent of applications, this is how many businesses view things: users think about performing a function in the CRM application without much awareness that they are participating in the order-to-cash process, and developers want a single application lifecycle to manage for both applications and related processes.

The opposite of BPMS vendors who have been building (often substandard) application development environments to extend their process modeling and execution, Agile Platform adds process modeling and execution to a robust application development environment. They have a proven track record in doing application development right; it remains to be seen if they’re doing BPM right as well.

ARIS Express Process Modeling Contest

The ARIS online community is running a contest until the end of November to get people started with process modeling: create a process model (using their free downloadable ARIS Express process modeler, of course) that falls into the category “BPM is useful” or “BPM is fun”, and submit it on their site. I’m not sure what you do if your process is both useful and fun. 🙂 The winner in each category will receive an iPod Touch.

You can also vote for your favorite model in each category. You’ll need to register as an ARIS community member to submit a model and to vote, but you can browse through the models without registration.

Giving Technology Back to the Community

I’m a strong believer that technology can be a way up for those in financially disadvantaged circumstances: without some computer skills, kids can’t compete in school, and don’t meet the minimum requirements for many jobs. One way that I can help – and probably many of you reading this – is to donate to programs that provide access to computers and training to people who can’t afford to buy them. There are a number of ways to do this: you can give money, you can give used computer equipment, you can give your time, and you can promote the programs to others who might do the same.

This week, I replaced my mother’s old computer, and was left with a working (although underpowered, by today’s standards) computer with keyboard and mouse. I immediately thought of Little Geeks, a program that refurbishes old computers, provides them for free to kids in need, along with 12 months of internet access and some training on how to use it. They use reBOOT Canada as their drop-off depot; reBOOT is a charitable organization that “provides computer hardware, training and technical service to other charities, non-profit organizations and individuals with limited access to technology”. I headed off to reBOOT yesterday to drop off the computer, and had a chat with Nicholas (I believe this was Nicholas Brinckman, the Executive Director). He mentioned that they’re trying to get funding from the Aviva Community Fund to build 50 learning centres across Canada, in partnership with community centres and schools.

If you support this idea, go to the reBOOT project page on the Aviva Community Fund site and vote for their project (registration required). You can vote once per day until this round of voting ends in 11 days, and I encourage you to drop in there daily to cast your vote if you believe that this is an important initiative. They make it easy to link to the page on Twitter and Facebook, so use your social network for good. You can also help out by dropping off your old computer equipment – and encouraging your employer to do the same when they sunset old computers, printers and other equipment – or volunteering some of your time to help with computer refurbishment.

Fujitsu Interstage BPM in the Cloud

In the Fujitsu briefing last week, I also heard about their cloud BPM offering. Interstage BPM has supported multitenancy for some time, allowing them to provide private BPM cloud infrastructure, most commonly used by business process outsourcing firms. Multitenancy is a key feature of true software as a service: a single software instance supports multiple clients by virtually partitioning the application and data, rather than setting up an independent instance of the software for each client.

Multitenancy is also key when you want to productize it on the web, since it allows for fast and easy provisioning of new accounts, and that’s exactly what Fujitsu is doing with the launch of InterstageBPM.com, which puts the full power of their BPM suite on the web. They have two free versions: a trial version allows for unlimited applications and process instances for five users, but only lasts for 30 days; and a team version, that allows for unlimited applications but only 250 process instances per month on an ongoing basis. Presumably, the team version is for developers, while the trial version is for a full production test or proof of concept. Above that is a single tier of paid licensing: $50/user/month for unlimited applications and 10,000 process instances per month. There’s another tier for solution providers, but pricing and details aren’t spelled out: that would more of a BPO or application development offering. All versions provide 99.88% availability – you’re not going to run your trading systems on this, but that’s fine for many human-facing business processes – and the paid enterprise version is supported by email and phone but currently only during US Pacific business hours. They also make it simple to move applications between the cloud and on-premise versions, similar to what Appian is doing, by providing an easy method to create an application package and move it between different instances and versions.

I find the cutoff of 10,000 process instances per month in the enterprise version interesting: that must be where Fujitsu feels the tradeoff is between cloud and on-premise systems. For smaller organizations, the usage model will likely be to use the free team version for development, then deploy on the enterprise version; larger organizations will more likely use either the team or enterprise edition for development, then deploy on premise. The cloud versions are also appropriate for third-party application developers, developing process applications on the Interstage platform that can be sold in the online marketplace to end-user organizations.

Aside from the usual arguments for cloud-based offerings, BPM in the cloud makes a lot of sense when you’re participating in processes that originate with multiple organizations. Having an RSS feed for any task list in the cloud-based BPM means that you can consolidate your view on multiple BPM instances from different organizations into your feed reader, for example, or push multiple feeds to a dashboard for monitoring. And although cloud-based BPM isn’t a prerequisite for large-scale federated processes, it can help to make things integrate more smoothly.

There’s a few reasons why this sort of offering makes sense from a company like Fujitsu. First, they’ve served the BPO market for quite a while, so they understand the practical issues of multitenancy in a way that few other BPMS vendors do. Second, this all runs on their own data centers, which also provide managed data center services for many customers in many countries. That means that they have a proven track record at keeping systems up and running, and they’re running on their own gear so there can be no finger-pointing in the event of a failure. Third, with 85 Fujitsu data centers around the world, they won’t be making the mistake of many other US-based cloud vendors by offering US-only data centers, which is an unacceptable solution to many non-US organizations (including all of my Canadian financial customers): although the initial version of their cloud offering is running in their US data center, they’ll be rolling it out to the others around the world.