I might be stretching a point a little (well actually quite a lot) with the title,  but just like Software as a Service gets rid of all the redundant effort that goes on in every enterprise delivering software,  so Architecture as a Service is doing the same thing.  A few examples of AaaS arrived in the RSS reader today:

1. Microsoft started a BETA process to design a set of documentation for the delivery of SoftGrid solutions and Server Virtualization,  hopefully by using the feedback processes associated with software beta’s they will be able to hone their documentation set collaboratively to the point where the amount of duplicate work within every enterprise is greatly minimized.  You can sign up for the beta at http://connect.microsoft.com/ using code IPDM-QX6H-7TTV
2. VMware have gone one better and setup a wiki to serve the same purpose, maybe more as it addresses the whole lifecycle from plan through to manage allowing the community to share everything right down to scripts, reports and tools, excellent!  You can find it here http://wwwa.vmware.com/www/wiki/index.php/Main_Page

Its also worth remembering that Microsoft have this cool technology in the works that allows you to model your infrastructure in Visual Studio and then provision it onto your utility computing infrastructure,  I wrote about it here.  And there are managed hosting alternatives that actually allow you to provision a whole virtual infrastructure on demand from a GUI development environment.

Now I know that architecture is really about developing a solution to meet business requirements,  which are often unique, but at least these initiatives and others are slowly chipping away at all that redundant work that goes on that doesn’t have to be unique, but is - just because there was no mechanism to share best practices across customers.

SharePoint 2007 strikes me as an excellent foundation upon which to build,  its extensibility seems impressive and the fact that it already includes basic versions of all the main elements, blogs, wiki’s, personal pages and people search provides them with a great learning platform.  Mike scores Microsoft pretty poorly so far:

• Blogs:B-
• Wikis:C+
• Tag/Social Bookmark System: N/A
• Social Networking: B+
• XML Syndication: N/A (feed aggregation and management)

but my point is that it’s just a matter of time,  Microsoft has all the technology infrastructure they need, all the research, all the resources and their B- existing infrastructure is giving them all of the practical experience (Microsoft thrives in practical experience). 

I am a little surprised that Microsoft haven’t purchased NewsGator though,  it seems a perfect complement to their ambitions (both within Windows/Office Live the enterprise) and and it’s all based on their technologies.  Makes me think they must have something up their sleeves.

It won’t be long though, I suspect we won’t even have to wait for a new version of SharePoint (but I have no inside information).

I have recently been thinking a bit about Enterprise RSS and getting frustrated by key features that seem to be missing, confusing or partially implemented, two examples being:

• Lack of unread status syncing between multiple clients of the Windows Common Feed Store
• Lack of support for basic authentication in the Windows Common Feed Store

I have also been thinking more broadly about security and management issues in general, so I was pleased to come across this post on the Collaborative Thinking blog which provided a useful list of topics to consider concerning enterprise RSS, my additions are in bold:

Security

• Ensure support for SSL and related authentication methods
• Ensure support for basic authentication over SSL as well as NTLM and Kerberos (which are mainly of interest in a Microsoft environment only)
• Some way to authenticate to RSS feeds that are protected by forms based authentication, common when enterprise portals are involved
• Browser add-ins should run “Internet zone” or “restricted sites zone”
• Require Windows XP SP2 Attachment Execution Service support
• Ask about block lists, content filtering, and handling of malformed feeds
• Outline how RSS enclosures are managed
• Require ACL’s per feed at group/user levels integrated via AD and LDAP
• Ability to encrypt the feed store and enclosures,  a lot of enterprise data is going to end up in this store, especially if enclosures are used
• I am also interested to see if some form of DRM could be applied to enclosures, with the digital rights acquired for the download user at the same time as the enclosure is downloaded.  I am concerned that enclosure enabled document libraries will result in very efficient document hoarding and resulting data leakage.  This may be a scenario that’s supported by Outlook and SharePoint 2007 IRM,  but I suspect that the common feed store DOES’ NT acquire the rights for the logged on user as part of the download process, which will make it difficult impossible to view the enclosures offline.

De-duping

• Assess how multiple updates to posts are handled and impact to read/unread marks
• Synchronization
• Check to make sure read/unread marks are managed across clients and devices

Storage

• Primarily a concern when feeds are sent into e-mail systems
• Potentially a compliance concern   

Network Utilization

• Assess download schedule management
• IE7 “salting” the interval (is it good enough)
• Is the RSS 2.0 Time To Live (TTL) tag honored
• Use of / support for Incremental feed updates (RFC 3229 (“Delta Encoding in HTTP”))
• Compression of feeds and enclosures

Management

• Ability to associate automatic feed subscription with LDAP directory group memberships, this really is going to be essential if feeds subscriptions becomes important to process compliance, e.g. a person gets informed when their Quality Management System is updated via a RSS feed.
• Ability to track usage

Personalized Feeds

• Not really an easy feature to implement at this time

Attensa and Newsgator seem to be addressing several of these issues,  but Microsoft and IBM don’t so it will be interesting to see how enterprises get on when they deploy Office System 2007 and Notes Hannover.

Brian Madden has a lot of useful comments on each feature, my favourites are in bold.

1. NTLM
2. kerberos
3. Digest
4. Forms based
5. Basic (usually combined with SSL)

Only the last of these mechanisms can be assumed (with any confidence) to work with most desktop RSS readers and often web based readers often don’t even support that. 

The essential issue is that the web pretty much assumes that we can cope with all of these authentication mechanisms because all access is interactive, but with RSS it needs to be automatic and transparent.  The following issues spring to mind:

1. NTLM – so far as I know is only supported by IE 7  and the Vista RSS platform and very few servers,  but if you are a Microsoft shop and are only connecting to SharePoint 2007 then this may work for you
2. Kerberos – same as above, but probably even more demanding
3. Digest – hardly ever used in my experience, either on client or servers, but I may be wrong and has the disadvantage that your username and password will need to be stored somewhere on the client, and many enterprise security policies don’t allow that
4. Forms based – very popular server side, but no chance of supporting this in everyday RSS readers.  I have seen a hack which involves browsing to a web page from within the RSS reader, authenticating, getting the cookie and then synchronising your feeds.   VERY VERY messy
5. Basic with SSL – very widely used, supported by most readers and RSS servers, but has the disadvantage that your username and password will need to be stored somewhere on the client, and many enterprise security policies don’t allow that

This leaves us with a problem.  If you are a Microsoft shop you might get away with a combination of 1,2 and 5.  If not then it looks like it’s time to start lobbying your security policy makers to allow Basic over SSL and local (encrypted) credential storage.

Now Rod thinks that these ‘Office’ applications will be displaced by tools like blogs, wikis and Microsoft’s Excel services.  I partially agree I think these tools will just complement the traditional Office applications and extend their reach beyond the desktop and the network file server.  Here are some (mostly Microsoft)examples:

1. Applications built by using custom searches, probably encoded in the URL, against Internet and intranet search engines.  Expect to see this particularly in situations where the search engines let you reach into databases and document metadata
2. Applications built using the incredible versatility of SharePoint custom lists and workflow
3. Applications built by combining InfoPath, Word, Excel XML documents with data selectively promoted from the XML into SharePoint lists
4. Applications built processing RSS data from all manner of applications and then mashing it up with other data, or pulling it into Office or web based analysis tools
5. Applications built using the next generation of web based 4GLs like DabbleDb and Coghead
6. Composite applications built by integrating portal components, including some or the above, like Intalio’s, built from Dabble DB, FeedBurner, FeedDigest Flickr, Google Analytics, LinkedIn, Technorati, WordPress,Zoho Writer.

This is the true long tail of software,  traditional office applications enhanced and integrated with web office, collaboration and line of business applications.  However lets not forget the humble desktop application.  Its had a bad reputation in the past – largely because of DLL hell – but as new classes of applications appear that don’t need to be installed in the traditional sense but can just be streamed down to the PC, cached and executed I expect that desktop apps will get a new lease of life in the enterprise. 

the right information, at the right time, in the right context, has been a holy grail for IT organizations for many years

In his article Mike describes 4 zones (my comments in red):

1. Salient Zone: Information directly relevant to the activity that has the user’s attention and primary focus, or fits a user profile that includes topics the person is interested in (with an implied immediacy in terms of awareness and delivery timeliness).   e.g. Project 1 – changes
2. Peripheral Zone: Information that is strongly-to-moderately associated to a set of activities that the user participates in or to their profile (exclusive of the current activity). While there is likely discretion in terms of how and when the user needs to be aware of the information, there is an implied desire for it to be readily “glanceable”.  e.g. Project 1 and Urgent
3. Ambient Zone: Information users should find interesting but could just as easily ignore. The information could be tertiary, having a no strongly patterned relationship to any activity. But it also might have some intriguing synergy with, or some discernable influence on, activities or other user interests. Communication here is more informal, with user no guarantee that users will divert attention and interact with the information.  Research
4. Nascent Zone:  information forming at its early stages that might have some latent relevancy in the future. Users might be interested in cycling through ever so often as part of general awareness and trend analysis.  Email and general feeds

The problem with the single inbox concept for information is that it provides all 4 types of information in a single stream, and it’s very difficult to cope with this by constantly scanning this stream.  Even when you apply methodologies like GTD, which force you to categorize email and other information it’s still difficult to see just the information you need in every particular context unless you are very organized.

RSS has a better chance than email because it arrives pre categorized, according to the feed title, which most aggregators use to deposit each feed in a separate folder.  Even better most aggregators allow groups of feeds to be placed in additional folders.  These groups of folders if structured around based on context will definitely help.

Consider the following folder structure:

• It allows me to click on Projects, and see all feeds from every project – useful for a daily activity scan.
• Click on Project 1 – changes and I am ready to focus on my Project change board meeting

The beauty of RSS is that once the structure is defined feeds look after themselves.  In addition in tools like Outlook it’s also possible to use flags and search folders to create lists of actions, items, folders with items aggregated by author, by date etc.

IBM are also moving in the direction of context zones with the context being an Activity (for me see here and here), and their activity explorer being the tool I interact with.  The activity concept is very powerful allowing me to associate RSS feeds, emails, IM messages, documents etc with an activity.

In addition there seems to be a general consensus amongst search gurus that search tools will soon be watching what we are doing, how busy we are, who we are talking too etc etc and will be presenting us – non intrusively but proactively - with the information we need from a wide variety of sources, transparently and without forcing us to interrupt we want to.

The best companies will figure out how to blend top down enabablement with bottom up productivity and innovation.  It’s great to see Gartner recognising this at last, in the past Gartner’s TCO model for PC’s has promoted the idea of “stop users fiddling with their desktop”  now they seem to recognise that at least for some users that fiddling was actually productivity tuning and process innovation!  Here is an encouraging quote from Jeffery Mann, a research VP at Gartner:

When I talk with customers about how to achieve a high-performance workplace (HPW), one of the hardest things for them to deal with is the need to loosen up on some control issues, and how to do that without losing control completely. This is natural. For the past several years, CEOs and CFOs have been asking CIOs to reduce costs, reduce risk, ensure compliance and generally take tighter control of users. This has resulted in locked-down desktops, strict TCO and ROI procedures, and tight IT procedures all around. The result is that IT has collectively become “The Abominable No Man”’ in many organizations, better at refusing or blocking any initiative than facilitating it.
We cannot stay on this trajectory. The complexity of the business and IT environments is too overwhelming to pursue the myth of total control. There are too many variables and influences to permit anyone to control all inputs. Even if we could, that would be a bad thing. Real innovation is coming from unexpected and not totally understood areas, such as Web 2.0 and consumer-oriented collaboration facilities. To block access to these is counterproductive and, ultimately, futile. Increasingly, many users see access restrictions as similar to network faults: a minor irritation to route around.

Of course it’s not about no control.  In my view it’s about IT progressively withdrawing to managing only those things that are business critical and enabling security and connectivity services, and even then considering whether they need to manage applications and data or whether they can get away with just controlling a standard web service, RSS feed, or email feed.  In Gartner’s words:

Does this mean we should throw open the doors to every virus-laden, spyware-filled download we can find? Or post sensitive information on any blog site we care to? Of course not. Loosening control does not mean giving up all control. It could mean enabling four or five different products in a particular technology area instead of just one (but not any). Innovative IT managers are experimenting with virtualization to shield experimental trials from sensitive corporate processes. In some cases, it will mean trusting employees to do the right thing, something businesses are accustomed to doing in other areas (like contract negotiations or travel expenses), but not done often enough within IT policies.

I like the idea that Jeff presents here of providing a choice of applications,  as this fits very well with my opening point about the different ways that people like to work.  As we see more applications that can interact with standard web services like RSS in a predictable way we will be able to move in this more flexible direction.

We want to be the single company to provide access to all applications, regardless of what type they are. For Windows client/server applications, you can use Citrix Presentation Server. For web applications, you can optimize them with NetScaler appliances. For full client or offline applications, you can use desktop streaming.

I like this positioning,  within my company I have been promoting a similar change in focus away from delivering technology services, more in the direction of application delivery, in fact I used the phrase:

Delivering capabilities to productive people and teams

which is pretty snappy, and goes beyond applications because I wanted to cover more than just applications, and include data, devices, voice etc into our scope and it also allows us to break down every term to provide a lot more detail.

I thought it would be useful to describe some of the characteristics of the newsgator approach:

• A loosely integrated client and services platform
• The services platform only asserts itself when it can add value to the user,  otherwise it gets out of the way and lets me get on with the job
• An eco-system of different client software optimised to address different platforms, user-types and user preferences, some supplied by newsgator but others just integrated with the services platform using their APIs
• Different installations of newsgator online enabled software are aware of activities on the other clients, for example my work client is aware of what I have already read or subscribed to at home
• Each installation of a client can optionally have a different configuration, so my work client could be configured differently to my home client if that’s what I wanted 
• A complementary web interface for when an optimised smart client is not available, or for when a web interface is the optimised solution
• All of my configuration data is held by the services platform and be easily extracted and moved some place else, in fact other systems can use it in place as it is all url addressable
• All (actually nearly all but they are moving in the right direction) configuration data is held by the services platform, so that once I have authenticated to it from a freshly installed client it seems like my environment again
• Higher performance than would be possible with a client only approach

In summary, the approach provides me with considerable freedom of choice, great performance, optimisation without sacrificing flexibility and openness, pretty inspiring!

Now if Microsoft can do the same thing by utilising Windows Live to make the experience of multiple personal PC’s, work and personal PCs, a household of PCs, PCs and Mobiles, multiple identities, sharing data, PC migrations and upgrades etc more seamless and deliver freedom of choice, great performance, optimisation without sacrificing flexibility and openness then I will be very impressed. 

It’s interesting that watching Ray Ozzie perform at the Financial Analyst meeting recently he definitely seems to have a newsgator-like vision for the eco-system of Microsoft products and services, and some of his recent innovations around the use of Live Clipboard and RSS and Simple List Extensions seem to indicate that he sees the value in delivering these innovations using standards so there may be hope.

The only thing I really need to round off the newsgator experience is for every bit of configuration data to persist server side (flags, snippets, saved searches etc) and applications that stream down to my new PC automatically with no possibility of conflicts with other applications and no administrative rights to run or install, for that I guess we need them to be .Net ClickOnce applications or wait for Microsoft to build Softgrid like virtualization support into their operating systems and Softgrid like streaming into their packaging approach and yes I do know that Microsoft has just acquired Softricity!

This would be a vision for Software as a Service that would really appeal to me!

With RSSBus, our goal is to offer a simple, easy alternative for the small organization with
little to no IT assets, little to no professional development tools, and no professional
programmers to use them.
If you spend much of your working day developing web services for one of the major
service platforms, and you can no longer write a single sentence without a four, five, or
six letter acronym in it, perhaps you should stop reading right here and spare yourself the
disappointment.

And if your job title has “Architect” or “Enterprise” in it, then please  stop reading here. 
Just in case you didn’t notice, we said Service Bus, and not Enterprise Service Bus.  In
fact, this is the last time we will mention Enterprise.

What we are building is something different, a service platform for the rest of us, the non-
acronym-speaking crowd.  If you have bits of pieces of data that you would like to quickly
exchange with and/or connect to other systems, if simplicity and ease of use is your most
important consideration, please read on.

Check out the list of pre-built modules and you will get a feel for the sort of application events that the developers expect RSS bus to act upon:

Here is an alphabetical list of some of the modules included in the product:
AmazonOps: operations that provide access to Amazon services.
BusOps: core Bus operations (feeds, caching, meta information).
CcOps: operations for credit card authorization and processing.
EbayOps: operations that provide access to EBay services.
FedexOps: operations for FedEx shipping and tracking.
FeedOps: operations for managing RSS feeds.
FileOps: operations for managing files and directories.
FtpOps: operations for transferring files to and from FTP servers.
GoogleOps: operations that provide access to Google services.
ImapOps: operations for receiving email messages from IMAP servers.
LdapOps: operations for connecting to LDAP directories
MediaOps: operations that provide information about digital media files.
OfxOps: operations for accessing bank accounts and financial services.
PaypalOps: operations that provide access to PayPal payment services.
PopOps: operations for receiving email messages from POP servers.
QbOps: operations that provide access to Intuit’s QuickBooks.
SearchOps: operations for searching and indexing.
SmsOps: operations for sending SMS text messages.
SmtpOps: operations for sending SMTP email.
SqlOps: operations for connecting to SQL databases.
LdapOps: operations for connecting to LDAP directories.
MediaOps: operations that provide information about digital media files.
OfxOps: operations for accessing bank accounts and financial services.
PaypalOps: operations that provide access to PayPal payment services.
PopOps: operations for receiving email messages from POP servers.
QbOps: operations that provide access to Intuit’s QuickBooks.
SearchOps: operations for searching and indexing.
SmsOps: operations for sending SMS text messages.
SmtpOps: operations for sending SMTP email.
SqlOps: operations for connecting to SQL databases. 

It’s not difficult to see that this solution is going to appeal to people who have grown up with the web, who love REST and Mashups etc, and it’s simplicity could well serve as an inspiration to many big EAI solution providers as well.

First try

• Terminal served into my desktop (2003 server remember) as admin
• Installed SVS
• Rebooted
• Tried to install Office 2007 in a layer, a layer is an SVS term that describes a way of isolating an application (by means of a file system redirector) from any dependencies on your PC and also isolating your PC from any changes that the application tries to make.  In other words – its safe!
• Install failed,  which is perhaps not surprising as I already have close to a hundred applications installed including OneNote 2007 and Altiris recommend installing on a clean machine.
• Started again

Second try

• Created an XP SP2 Virtual machine using VMware 5.5
• Installed all patches and VMware tools
• Installed SVS
• Rebooted
• Installed Office 2007 into a layer, worked fine
• Exported the layer to network drive (failed)
• Exported the layer to the VMware Virtual drive (worked)
• Copied the exported file to network drive
• Terminal served into my desktop (2003 server remember) as admin
• Imported the exported file into a layer on my desktop
• Activated the layer
• Went back to my normal user account
• Double clicked a PowerPoint file,  PowerPoint 2007 opened and ran fine (very limited testing)
• Clicked on PowerPoint 12 icon in Start Menu, worked fine (very limited testing)
• Went back to my admin account
• Deactivated the layer
• Double clicked on a PowerPoint file, PowerPoint 2003 opened
• Note: In the above activate and deactivate actions I did not need to log-off or reboot

So I now have Office 2003 installed on my desktop, and office 2007 available as a layer that I can activate as required.  I am impressed enough that next time I rebuild my desktop,  I will probably install all applications as layers,  although except for testing out new applications I don’t think I will use SVS extensively until then.  I will also look forward to some admin utilities being developed that allow me to copy files around between machines or do bulk imports and activates, so that maintaining multiple machines and rebuilding them becomes less of a chore.  Of course Altiris has enterprise scale tools to do this,  but I only have 9 physical PC’s on my home network