We open with video, quite long. It’s two parts. The first being a community thanks and embracing of the community. Perhaps a little cheesy, but I do like that executives at Microsoft realize this isn’t just a conference where workers show up and learn. It’s a closer knit community.
The second part of the video is a commercial. It’s a lot of sound bites and comments from people at Microsoft and customers. It starts out with a bit of cloud push, talking Azure, talking hybrid, and then going into more “Microsoft is great” comments.
I think think this is redundant. We’re all Microsoft devotees. We’re sold. Personally I dislike the “sales” of the platform when we’ve already decided to purchase the platform and build careers on it.
T. K. “Ranga” Rengarajan is up first. He’s an executive in the CVP Data Platform.
We have an explosion of devices. 1.3B in 2010, projected to grow to 25B in 2020. That’s amazing, and that makes a different. However, it’s data growth, projected to be 40% year over year, that matters.
40%.
If that happens, that’s incredible. I personally think that’s high, but perhaps not. Certainly unstructured data growth is crazy. Now, can Microsoft enhance and grow Filestream/Filetable/Full Text Search to keep up? We’ll see. The last version was disappointing here.
Overall, some good stats, but I think this keynote makes a few mistakes that I’ve made. It preaches to the choir, talking about things we also believe and know. It’s not inspiring or driving us to move in a direction.
We get a shot of Microsoft isn’t like other vendors. We can do anything.
Huh? That’s a bit of a silly statement. And somewhat arrogant. It feels contrived to me.
Ranga mentions that they’re “proud to be schema free with DocumentDB”. It’s a statement at the end of Ranga talking about Azure DocumentDB and how all these devices that need to capture and use data, need a way to do so. I’m not sure that I like the idea of schema free.
More cloud, more ne wfeatures in Azure, some of which were announced last week at TechEd in Barcelona. I think some of these are good features, niche-y, but useful for people.
SQLDB – The scale out approach, the practical approach. They want to be simple, but also powerful. There’s new SKUs, up to 640 cores, 64 virtual CPUs, 1TB of RAM. I’m sure that will help some, but if they price is what I think it is, can many justify it? I doubt it,
A demo from the Pier One Imports site. Showing some auto-fill/auto correct on the form from partial entries, or misspelling. We see some code on how to get this easily into your apps. That’s good, though to be fair, I’m not sure that’s as much a DBA/data pro thing as it is a developer feature. Good, but perhaps mis-aimed for the audience.
Demo 2 – DocumentDB, quick loads of JSON data. Less than impressive.
We then get a bit of a sharding demo, with some code on how this is implemented, seeing ADO.NET code to help with sharding and auto growth of instances, but it’s a bit underwhelming. I find myself less than excited by this, partially because the story isn’t told well. Watch for yourself and decide.
We see a few clicks to geo-replicate databases across the world. Not a great demo, but they have made it much easier to get copies around the world if you have the need. No idea on cost, limitations, but it’s still a feature that is needed.
“”Simplify with cloud” – Take any data, in any format, and put it in the cloud. Then transform it and then get answers. I’m not sure that’s my world. At least not in a company and at scale. Far, far, too much of my data and work needs to happen outside of the cloud.
A few notes from DELL IT, which was lots of buzzwords that didn’t provide any details or really any information. I’m disappointed.
Major update announcement, new features coming to Azure SQL DB, but coming later this year, and in preview.
I’m not sure what future items, which aren’t available, are major announcements. This reminds me of the vaporware announcements of so many companies in the past.
A live demo, which is cool. As Jen Stirrup noted, no worry from the presenter, which shows some confidence in what they did. A site where we can click on one of three items to purchase something. Try it at http:aka.ms/gopass. People are clicking on devices and in real time we see the totals being aggregated on screen, with Perfmon showing transactions/sec.
Then we see this is in-memory OLTP, but with a writeable columnstore index. That’s cool.
Database Stretch – We see part of our database in Azure, and part on premise. I’m sure it’s not as simple as it seems, but it’s a cool concept. We also see queries being spread out to Azure, combined with local on-premise database. We also see a restore, in real time. That’s cool, though no idea on how the costs might cause issues.
Joseph Sirosh, VP of machine learning and something else at MS. He’s come from Amazon and shows a nice introduction. I do think this is the kind of stuff that is different in the SQL Server community. I don’t think any other conference, executives, companies, ever would have put up an intro like this in years past, or even today at many other tech conferences.
We are more than work, and it’s good to see that on stage. Whether it’s a play or not, it’s a recognition of what we are.
We now start to see the Azure Machine Learning introduction. The description is too vague, too full of buzzwords. However he knows this is the case, and promises demos to show how this works.
Machine Learning – An interesting demo. We see Pier1, a store, using Kinect sensors to generate a heat map of where people are moving in the store. No real personal details, but just a map of density. That could be valuable, but the idea of re-arranging the “end caps” on aisles in response to movement is a bit contrived. This is the type of “hyper” change that doesn’t always produce the results you want. The world isn’t that simple.
We move to customized searching and emails for customers, based on machine learning. The presentation that talks about the machine knowing what you want is slightly creepy. Seeing a Cortana demonstration integrated into the Pier1 app is interesting. I don’t know how many people use that for search and help, but if there is any significant number, then it’s a good idea to experiment with.
Minus the creepy “the machine knows you” messaging.
Certainly we need better machine learning that tries to use more data to target people in an effective manner than the simple “what you did last time”, or “this is what others did” recommendations we get. The fear from me is that marketing people won’t get that we also want “fewer” targets, not better ones.
Next we have a look at Power Query and Power Pivot. Lots of downloads from people. One great quote “Data is a bucketful of potential”. I really agree.
Let’s see BI, a live dashboard for a Pier1 store manager. We have a PowerBI system with tiles that we can drill through and get more information. It’s pretty, with lots of information. I think this stuff can be useful, but I also think it needs some training and work from end users. I wonder if the constant changing of tools helps end users get better or more confused?
I’m not knocking what Microsoft is doing. I do think that if they can make things easier for users that’s good. However I also think that much of this complexity and the additional options can overload people. I sometimes wonder how much people really understand about what they’re seeing.
Not a great keynote, disappointing. I wish that we’d stop getting executives and program managers that aren’t good speakers. It’s really not making a difference. Hire an actor and use them.
Voice of the DBA 