Realtime predictive intelligence

Momentum Berlin 2011 has ended. It was a great, well organised event. The only point of critic I can give is to create more opportunities to engange in networking with other people. Don’t get me wrong, the parties (ewerk and 40 seconds) were great but the music was a little bit too loud to have a good conversation. Nonetheless, we did have a lot of fun!

So what did we learn this year? Apart from the C6 acquisition, no real surprises. Here’s a small summary of what was new for me.

xMS

xMS is new deployment model. The idea is to create a blueprint of your Documentum project and to deploy this blueprint to a VMWare environment without the need of someone installing and configuring everything manually. For larger projects this could mean that you can reduce the time to deploy from days or weeks to just a few hours.

We saw two great examples of how EMC wants to become more intelligent in its content delivery, improving user experience a lot. BOCS is great when having branch offices but content would only get cached if it was accessed first.  The plan is to predict what kind of content users in a certain branch office access a lot and pre-cache similar data. On top of that BOCS will also allow async writes, allowing you to write to BOCS first, who will then asynchronously write content to the content server. You all probably know Amazon’s feature “People who bought this book also bought …”. xPlore will be able to do something similar (We saw a demo in centerstage).

xPlore

There is more to say about xPlore. There will be query based subscriptions, thesaurus support and EMC continues to improve performance as well.

The unified client

EMC announced the acquisition of C6 technology. Are we happy with that? I heard a lot of positive reactions but EMC really has to sort some things out as not everything I heard was positive … Personally I have mixed feelings. For me, D2 is webtop, and X3 is centerstage, based on a common configuration layer.  The common configuration layer is a good thing but you have to be very careful with that. It sits on top of DFC or DFS which means that from an architectural point of view you absolutely cannot implement business logic in that configuration layer or you’re in trouble with other applications. If you do that, you lose all the power of TBO. We know C6 for quite a few years now in Belgium and D2 is not new to me. It’s not a major success here because of what I mentioned earlier, it really just looks like webtop and in terms of customization you’ll hit a wall much more quicker than you will with WDK.

Press release acquisition C6 technology

D7

Documentum 7 is announced for Q3 2012. No major surprises here. In D7 there will be a major performance increasement regarding to creating sessions and session management. EMC’s focus is changing to Restful services and will fully support all services as Restful. DFS will become the primary interface instead of DFC. D7 will also come with extensive monitoring options, something a lot of people will applaud. I’m also looking forward to the content server “Dormant state” which actually sets the content server to read only, ideal for patch upgrades without downtime for consumers.

Conclusion

For the first time since long I have a good feeling about Documentum long term and Jeroen Van Rotterdam plays a big role in that. He has a clear vision and there is a lot of innovation going on. On the short term there is still a lot of confusion about EMC’s UI strategy.

See you next year in Vienna!

A few weeks ago I finished all of my basic trainings of McLaren Software: Business Fundamentals, Configuration and System Administration. Back in July I posted my first impression on Work Manager after my first hands-on experience. Now I’ve used it, customized and tweaked it. Time for another review.

Work Manager is a product that offers a lot of things out of the box. This is a very important aspect for decision takers in ECM. In theory this means less maintenance and upgrade costs, reducing the TCO. This is true but I believe in the fact that sometimes small tweaks in how things work can be a big added value for the product. A lot of customizations are of great benefit to the end users and after all, it’s the end user we need to please in a succesful project. Who’s using webtop completely out of the box without any configuration or customization? To be short, configuration and customization is important, and spoken out of experience, a lot of buy/no buy decisions are taken based on that fact as well.

So, how about Work Manager (Enterprise Engineer)? Work Manager is a fantastic product to tweak according to your needs. Roughly estimated, 95% of common adjustments can be done by configuration. During my test I also got the impression that the product is very mature which makes me feel confident about a coming project with Work Manager. I’ve seen EPFM into action but while EPFM is a set of services, integrated within webtop offered by EMC, Work Manager is a real product with a clear roadmap and a future. (EPFM is based on WDK)

Work Manager takes some time to start up but is in my opinion still reasonable for a large application with a rich user interface. Outlook or Word does not pop on your screen in less than a second either (At least not on my system). A small drawback for me is the fact that McLaren has put some business logic in its webservice layer. They had to, due to the fact that they also want to support FileNet with exactly the same set of features. Not a big issue and off course you can still use native Documentum as is together with Work Manager.

Momentum 2011

Days before Momentum 2011 starts in Berlin, there’s a lot to say about EMC and Documentum. License earnings in Q3 were lower than the year before – third quarter in a row this year, Gartner placed EMC far from Microsoft, IBM and Oracle in its new Magic Quadrant of Enterprise Content Management and Oracle is promising free WebCenter licenses for companies moving away from Documentum.

Earnings

EMC’s third quartner earnings were good, apart from one particular segment: IIG. For the third time this year, IIG’s quarter earnings were lower than the year before.  No surprise if you ask me. Documentum’s backend system is considered top of the game for many years. Up till a few years ago EMC’s competitors were far off nor did they have any appealing client interfaces. Now they do …

“Revenue from our Information Intelligence Group was $171 million, flat from Q2. IIG continues to transition its offerings to lighter-weight, content-enabled applications on top of xCP using modern virtualized frameworks. In Q3, we announced Documentum clients for iPad, an app which provides users secure access to information and Documentum and enables them to access share and collaborate on their preferred device.”

I’m sorry EMC but this is not convincing either. You’ve worked way too long on a product like Centerstage before it was considered ready for production. You really need to up your stakes or change strategy.

Gartner

Last year EMC was located very close to Microsoft, Oracle, IBM and OpenText in the Magic Quadrant of Gartner. Really difficult to pick a winner. Now, one year later EMC suddenly no longer looks like a real contender to Microsoft Oracle and IBM. Why? Difficult to say. What is important for Gartner and what is not? For my feeling Oracle and Microsoft are slightly better in web content management but there’s still a gap in document or record management compared to Documentum. If someone knows Gartner’s criteria, please let me know.

Oracle

Trade in your Documentum licenses and get Oracle WebCenter for free. What? WebCenter is an enterprise portal that can be used to display information about people, processes and data. I don’t know much about WebCenter but I doubt one can replace Documentum with Oracle’s product. It’s like comparing apples to oranges. You’ll need more that just the WebCenter product to replace Documentum. First impression? A marketing stunt.

Link: Oracle WebCenter blogpost

Update: It looks like more people are sceptical about Oracle’s offer

It’s Momentum next week so I’ll hope to learn more about the direction EMC wants Documentum to go.

Be careful with .GetBuffer() on a MemoryStream in .NET

Watch out when using the .GetBuffer function on a MemoryStream in .NET. Recently we did a change in an application where we query an inbox for emails with attachments. Those attachments used to be saved to a file first and then send over the network as a base64 string. With the change we no longer save to a filestream but to a memorystream, skipping the unnecessary IO.

All went fine until we received a complaint that a user couldn’t open an extracted attachment. The file appeared to be corrupt. After some investigation we noticed that all of the attachments had a strange file size: 8kb, 16kb, 32kb, 64kb, … What happened? We used .GetBuffer to get the byte array of a memorystream. A memorystream automatically allocates a buffer and when using .GetBuffer you also retrieve the unused bytes. If you only need the used bytes you have to use the ToArray() function.

Strangely enough most of the documents (PDF) opened well in Acrobat Reader, even with the unnecessary/empty bytes.  Only when the empty buffer at the back of the file grew too large, we got an error. After trimming all of the unnecessary bytes at the back of these files all opened fine again.

MSDN describes this behaviour: http://msdn.microsoft.com/en-us/library/system.io.memorystream.getbuffer.aspx

SAP Consuming a .NET Webservice containing an “AnyType”

We had a few problems with SAP consuming a .NET webservice. The problem was that the post contained an array of “Attribute” objects. This Attribute object consists of 2 fields

• Name field: a string
• Value field: could be anything (an “Object” type in .NET)

The WSDL of this Attribute object looks like this

<s:complexType name="Attribute">

    <s:sequence>

        <s:element minOccurs=”0″ maxOccurs=”1″ name=”Name” type=”s:string”/>

        <s:element minOccurs=”0″ maxOccurs=”1″ name=”Value”/>

    </s:sequence>

</s:complexType>

There were 2 problems:

The SAP proxy

The SAP proxy that is generated cannot handle this untyped element. Someone had to manually adjust the SAP proxy code so that they could define the <Value> node in XML markup in the SOAP message manually.  Unfortunately I’m not a SAP expert so I can’t tell you exactly how to modify the proxy.

The Value node

The second problem is the value node itself. The real problem is that SAP does strange things with the XML namespaces when sending out the SOAP message. It does not define the namespace of the webservice at some top level but redefines is at almost every node making it a) much more complicated than it really is and b) create a lot more network traffic because of all the unnecessary bytes being sent over the network.

An example:

               <n0:Attribute xmlns:n0="http://mynamespace">
                    <n0:Name>some string array</n0:Name>
                    <!-- The Value node is manually generated in the SAP proxy -->
                    <n0:Value asx:root="Value" asx:nsd="xsd wsd" 
                              type="ArrayOfString" 
                              xmlns:asx="http://www.sap.com/abapxml">
                        <string>first string</string>
                    </n0:Value>
                </n0:Attribute>

The value node is a string array, defined as the ArrayOfString complex type in de WSDL of the webservice. (Note the n0 namespace that is added on every Attribute node and the additional SAP namespace). The above didn’t work because the .NET webservice incorrectly deserialized the Value node

• type is an attribute in the http://www.w3.org/2001/XMLSchema-instance namespace. This namespace is not defined at any top level in the generated SOAP message and is redefined every time it’s needed
• ArrayOfString is a complex type in the namespace of our webservice http://mynamespace

Eventually the following did work

               <n0:Attribute xmlns:n0="http://mynamespace">
                    <n0:Name>some string array</n0:Name>
                    <!-- The Value node is manually generated in the SAP proxy -->
                    <n0:Value xmlns="http://mynamespace" asx:root="Value" asx:nsd="xsd wsd" 
                              xsi:type="ArrayOfString" 
                              xmlns:asx="http://www.sap.com/abapxml
 xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
                        <string>first string</string>
                    </n0:Value>
                </n0:Attribute>

This was our solution. If someone knows a better one, please let me know!