Tag Archives: cloud computing

SQL Server 2014 is done: Hekaton, Azure integration

March 18, 2014 onlyconnect

Microsoft has released SQL Server 2014 to manufacturing (an odd phrase in these diskless days) but which signifies that it is code complete for the initial release. General availability is April 1st.

What do you do if hardware trends enable you to stuff vast amounts of RAM into your server, along with many CPU cores? The answer is that you optimize applications to work mostly in RAM, with disk important as a persistence layer. This contrasts to the approach when you have large amounts of disk space and little RAM, when you focus on loading only as much data into memory as you absolutely need.

The implications for a database server are profound. Instead of a logic that goes something like “read from disk, do something, write to disk” you can address the data directly; it is just a memory pointer.

Now combine that with stored procedures compiled to native code. Performance leaps up, and by much more than you get simply by caching data in RAM, or using fast SSD storage, but still using the old disk-based approach in the database engine.

This is the reasoning behind “Hekaton”, properly known as In-Memory OLTP (online transaction processing), which is a new in-memory database engine that comes with SQL Server 2014.

It is fully integrated. You just have to add a filegroup to a a SQL Server database with the keyword CONTAINS MEMORY_OPTIMIZED_DATA and then create a table with the keyword WITH (MEMORY_OPTIMIZED=ON). And for the stored procedures, use WITH NATIVE_COMPILATION.

The speed-up is as great as you would expect. I have seen demonstrations of 30x or more performance increases, like this one in a demo based on one from the SQL Pass conference, but which I did for myself in one of Microsoft’s “Hands On Labs”:

In another demo, on an Azure VM, I got a speed up of 7x. Only seven times faster! Still, hard to complain about those sorts of numbers.

Unfortunately, in-memory OLTP is spoilt by some rather severe limitations in this release. The first problem is that a combination of the need to support native compilation of stored procedures, and other limitations, means that only a subset of T-SQL (the query and management language of SQL Server) is supported. You can see the list of what is not supported here; and it is depressing reading, with lots of keywords that you likely do use at the moment; even IDENTITY is on the list of what does not work.

Another issue is that the ability of In-Memory OLTP to take advantage of hardware is not as extensive as you might hope. Lead program manager Kevin Liu told me at a recent press workshop that the team recommends restricting total data size to 256GB, and that the recommended number of CPU sockets is two. You can get servers today with much more memory and more sockets. It gets complicated though: in a multi-socket server memory has processor affinity and there is a thing called NUMA (Non-Uniform Nemory Access) that describes the way memory is shared between processors.

According to Liu, Microsoft expects to lift these limitations in future releases, as well as improving T-SQL support, but things like this remind you that it is a version one release.

What else is in SQL Server 2014? There is some neat Azure integration, including a managed backup tool that is almost one click to have your data backed up to Azure storage; a brilliant facility for small businesses. You can also use Azure for high availability, creating always-on replicas in Azure VMs.

Data warehouse users will like the new clustered columnstore indexes, which allow you do use a column-oriented table structure for much faster processing of typical report and analysis queries. Columnstore indexes first appeared in SQL Server 2012 but were not updateable. Now they are.

SQL Server is well liked, licensing hassles aside; and even on licensing, Microsoft can always point at Oracle and claim, rightly, to be cheaper and less complex. It has earned a reputation for solid performance. SQL Server 2014 looks as good as ever, even if the management tools now look rather dated – the shell for SQL Server Management Studio uses an old version of Visual Studio, which is one of the reasons. I also suspect the SQL Server team lacks a dialog designer, but doubt that the average database admin cares one jot.

That said, it is difficult to describe this as a must-have upgrade, unless you can make good use of “Hekaton” in-memory OLTP. The porting effort will be worth it presuming you can get it to work. One of the good fits for the technology is managing web app session data, or, as in the example above, rapid processing to display recommendations or customisations on a web site.

I can imaging though that many users will look at Hekaton and decide that it is too much work or too immature for immediate use. What is left for them, apart from some nice Azure integration?

Not a huge amount, it seems to me, making this to my mind a transitional release.

Are you planning to upgrade? I would be interested to know your reasons why or why not.

cloud computing, microsoft

OneDrive, SkyDrive, whatever: Microsoft needs to make it better – especially in Office 365

January 29, 2014 onlyconnect

This week brought the news that SkyDrive is to be renamed OneDrive:

For current users of either SkyDrive or SkyDrive Pro, you’re all set. The service will continue to operate as you expect and all of your content will be available on OneDrive and OneDrive for Business respectively as the new name is rolled out across the portfolio.

I have no strong views on whether OneDrive or SkyDrive is a better name (the reason for the change was a legal challenge from the UK’s BSkyB).

I do have views on ~~SkyDrive~~ OneDrive though.

First, it is confusing that OneDrive and OneDrive for Business share the same name. I have been told by Microsoft that they are completely different platforms. OneDrive is the consumer offering, and OneDrive for Business is hosted SharePoint in Office 365. It is this paid offering that interests me most in a business context.

SharePoint is, well, SharePoint, and it seems fairly solid even though it is slow and over-complex. The Office Web Apps are rather good. The client integration is substandard though. A few specifics:

Yesterday I assisted a small business which has upgraded to full-fat Office 365, complete with subscription to the Office 2013 Windows applications. We set up the team site and created a folder, and used the Open in Explorer feature for convenient access in Windows. Next, run Word, type a new document, choose Save As, and attempt to save to that folder.

Word thought for a long time, then popped up a password dialog (Microsoft seems to love these password dialogs, which pop up from time to time no matter how many times you check Remember Me). Entered the correct credentials, it thought for a bit then prompted again, this time with a CAPTCHA added as a further annoyance. Eventually we hit cancel out of frustration, and lo, the document was saved correctly after all.

Another time and it might work perfectly, but I have seen too many of these kinds of problems to believe that it was a one-off.

Microsoft offers another option, which is called ~~SkyDrive~~ OneDrive Pro. This is our old friend Groove, also once known as Microsoft SharePoint Workspace 2010, but now revamped to integrate with Explorer. This guy is a sync engine, whereas “Open in Explorer” uses WebDAV.S

Synchronisation has its place, especially if you want to work offline, but unfortunately SkyDrive Pro is just not reliable. All the businesses I know that have attempted to use it in anger, gave up. They get endless upload errors that are hard to resolve, from the notorious Office Upload Center. The recommended fix is to “clear the cache”, ie wipe and start again, with no clarity about whether work may be lost. Avoid.

One of the odd things is that there seems to be a sync element even if you are NOT using SkyDrive Pro. The Upload Center manages a local cache. Potentially that could be a good thing, if it meant fast document saving and seamless online/offline use. Instead though, Microsoft seems to have implemented it for the worst of every world. You get long delays and sign-in problems when saving, sometimes, as well as cache issues like apparently successful saves followed by upload failures.

OK, let’s use an iPad instead. There is an app called SkyDrive Pro which lets you access your Office 365 documents. It is more or less OK unless you want to share a document – one of the the main reasons to use a cloud service. There is no way to access a folder someone else has shared in SkyDrive Pro on an iPad, nor can you access the Team Site which is designed for sharing documents in Office 365. Is Microsoft serious about supporting iPad users?

Office 365 is strategic for Microsoft, and SharePoint is its most important feature after Exchange. The customers are there; but with so many frustrations in trying to use Office 365 SharePoint clients other than the browser, it will not be surprising if many of them turn to other solutions.

cloud computing, microsoft, windows

Microsoft financials: record revenue, consumer sales declining in drift towards Enterprise

January 23, 2014 onlyconnect 2 Comments

Microsoft has announced record revenue for its second financial quarter, October-December 2013. Revenue was bumped up by the launch of Xbox One (3.9 million sold) and new Surface hardware. The real stars though were the server products:

SQL Server continued to gain market share with revenue growing double-digits.
System Center showed continued strength with double-digit revenue growth.
Commercial cloud services revenue more than doubled.
Office 365 commercial seats and Azure customers both grew triple-digits.

says the press release.

Another plus point is Bing, which Microsoft says now has 18.2% market share in the USA. Search advertising revenue is up 34%.

It is not all good news. While Microsoft is doing fine in server and cloud, the consumer market is not going well, leaving aside the expected boost from a new Xbox launch:

Windows OEM non-pro revenue down 20% year on year (that’s consumer PCs)
Office consumer revenue down 24% year on year – partly attributed to the shift towards subscription sales of Office 365 Home Premium

As usual, I have put the results into a quick table for easier viewing:

Quarter ending December 31st 2013 vs quarter ending December 31st 2012, $millions

Segment	Revenue	Change	Gross margin	Change
Devices and Consumer Licensing	5384	-319	4978	-153
Devices and Consumer Hardware	4729	+1921	411	-351
Devices and Consumer Other	1793	-206	431	-455
Commercial Licensing	10888	+753	10077	+751
Commercial Other	1780	+391	415	+199

The categories are opaque so here is a quick summary:

Devices and Consumer Licensing: non-volume and non-subscription licensing of Windows, Office, Windows Phone, and “ related patent licensing; and certain other patent licensing revenue” – all those Android royalties?

Devices and Consumer Hardware: the Xbox 360, Xbox Live subscriptions, Surface, and Microsoft PC accessories.

Devices and Consumer Other: Resale, including Windows Store, Xbox Live transactions (other than subscriptions), Windows Phone Marketplace; search advertising; display advertising; Office 365 Home Premium subscriptions; Microsoft Studios (games), retail stores.

Commercial Licensing: server products, including Windows Server, Microsoft SQL Server, Visual Studio, System Center, and Windows Embedded; volume licensing of Windows, Office, Exchange, SharePoint, and Lync; Microsoft Dynamics business solutions, excluding Dynamics CRM Online; Skype.

Commercial Other: Enterprise Services, including support and consulting; Office 365 (excluding Office 365 Home Premium), other Microsoft Office online offerings, and Dynamics CRM Online; Windows Azure.

Here is what is notable. Looking at these figures, Microsoft’s cash cow is obvious: licensing server products, Windows and Office to businesses, which is profitable almost to the point of disgrace: gross margin $million 10,077 on sales of $million 10,888. Microsoft breaks this down a little. Hyper-V has gained 5 points of share, it says, and Windows volume licensing is up 10%.

Cloud (Office 365, Azure, Dynamics CRM online) may be growing strongly, but it is a sideshow relative to the on-premises licensing.

How do we reconcile yet another bumper quarter with the Microsoft/Windows is dead meme? The answer is that it is not dead yet, but the shift away from the consumer market and the deep dependency on on-premises licensing are long-term concerns. Microsoft remains vulnerable to disruption from cheap and easy to maintain clients like Google’s Chromebook, tied to non-Microsoft cloud services.

Nevertheless, these figures do show that, for the moment at least, Microsoft can continue to thrive despite the declining PC market, more so that most of its hardware partners.

Postscript: Microsoft’s segments disguise the reality of its gross margins. The cost of “licensing” is small but it is obvious from its figures that Microsoft is not including all the costs of creating and maintaining the products being licensed. If we look at the figures from a year ago, for example, Microsoft reported a gross margin of $million 2121 on revenue of $million 5186 for Server and Tools. That information is no longer provided and as far as I can tell, we can only guess at the cost per segment of its software products . However, looking at the income statements, you can see that overall Microsoft spent $million 2748 on Research and Development, $million 4283 on Sales and Marketing, and $million 1235 on General and administrative in the quarter.

amazon, cloud computing, microsoft

Google Compute Engine: good enough to take on Amazon?

December 10, 2013 onlyconnect

A week ago, Google make its Compute Engine generally available. The service offers virtual machine instances as a cloud service, at prices from $0.114 per hour for a single-core VM with 3.75 GB RAM. In addition, you pay for outgoing network traffic and persistent storage. Reflecting the shortage of IP addresses, a static IP costs $0.01 per hour – but only if it is not in use. Linux is the only available operating system.

The service seems similar to Amazon’s Elastic Compute Cloud (EC2), but there are a couple of reasons why Google has the potential to take on Amazon. One is that it has the scale: just as Amazon, prior to the launch of EC2, had datacenters already in place to run its ecommerce business, Google has them to run its search and advertising business, as well as services like the Android Play Store, Google Mail, Docs and other cloud services.

Second, Google can afford Amazon-like commodity pricing. It could even afford to lose money on cloud hosting for an extended period, thanks to its dominance in web advertising, if it needed to do so in order to win market share (though I am not suggesting that it is in that position).

Why though would anyone use Google rather than Amazon? A post on Quora highlights some of the reasons, including sub-hour billing, live migration of VMs (no downtime), persistent disks that can be mounted read-only by multiple VMs, more integrated virtual networking, and better network throughput. This last point is interesting: the suggestion is that Google can use its own private connections between datacenters, where Amazon is more dependent on the public internet.

Amazon also has advantages, including a larger portfolio of cloud computing infrastructure services thanks to its greater maturity. Unlike Google Compute Engine, Amazon supports Windows VMs, for example.

Some large customers will want to spread VMs across multiple cloud providers for resilience, and it will not surprise me if Amazon plus Google becomes a popular combination.

development, mobile, salesforce.com, software development

Making sense of Salesforce 1 (it’s all about mobile)

November 20, 2013 onlyconnect 1 Comment

At its Dreamforce conference in San Francisco, Salesforce has been hyping up its newly announced Salesforce 1. The keynote left us in do doubt: it is fantastic, it does mobile, it does cloud, it does “internet of things”.

Co-founder Parker Harris describes Salesforce 1 at Dreamforce

But what is Salesforce 1? For those of us who like fluff-free facts, it has been difficult to discern. The APIs that make up the Salesforce 1 platform seemed on the face of it to be the same ones Salesforce has always had; yet the company says it has multiplied the number of APIs by 10 to create Salesforce 1 (a figure I still find hard to understand).

It is beginning to make sense to me. Salesforce 1 is a brand, a platform and an app.

As a brand, Salesforce 1 encompasses all the APIs that form the Salesforce platform. The best place to understand the current state of Salesforce 1 is here, where you can see links to all the APIs, including Force.com, Heroku, ExactTarget, Radian6 (social media listening), Pardot (sales automation), Desk.com (service cloud) and GoInstant (build real-time multi-user apps). Those individual APIs still exist in their own right, but Salesforce 1 is a new brand that encompasses all of them.

There is also a Salesforce 1 app for iOS and Android. This is mainly an HTML5 app, which makes it odd that it is iOS and Android only. As I understand it, you can also use a mobile browser and get a similar experience, so it might not be too bad for Windows Phone users after all.

The Salesforce 1 app is actually an evolution of the Chatter mobile app. As I understand it, it is built with the Aura framework, for creating a responsive user interface, with strong support for touch control. The Chatter app was renamed Salesforce 1 at the start of Dreamforce.

The Salesforce 1 app is built around a feed, and Salesforce describes it as a feed-first approach. Chatter has support for Publisher Actions, which now in Salesforce 1 have a more prominent role, making the feed capable of initiating tasks and being a mobile-friendly centre of operations. Some vendors I have spoken to, such as FinancialForce (wholly owned by Salesforce), see this feed-first approach as being the core of what Salesforce 1 is about.

When Salesforce talks about creating Salesforce 1 apps, that might refer to either of two things.

One is to create custom apps for your Salesforce users, which you can do without needing much code in some cases, which will be viewed through the Salesforce 1 app.

The other is to use the Mobile SDK for iOS or Android to create a native app. This does not have to be an HTML5 app, but could be if you want the quickest route to something that works.

According to CEO Marc Benioff, speaking to the press, much of the effort behind Salesforce 1 was in making the Salesforce browser UI properly mobile-friendly. He said that this includes mobile client libraries as well as the server APIs. Salesforce has an rapid visual builder for browser apps running on its platform, called VisualForce, and apparently getting these apps working nicely on mobile took huge effort.

Benioff gave the impression that VisualForce now works perfectly on mobile, but the booklet given to developers expresses reservations:

Only VisualForce pages enabled for Salesforce Mobile Apps and attached to a tab can be added to the Salesforce 1 navigation menu. Note that you may have to optimize these pages to work and/or display correctly on a mobile device.

Nevertheless, you can see the intent here, that anything running on Salesforce will work well on a mobile device. Benioff says that he only takes a smartphone with him when travelling, no laptop or even tablet, and he expects to be able to do all his work through it.

You could therefore call Salesforce 1 the optimisation of the Salesforce platform for mobile, subject to the iOS/Android limitation.

According to Salesforce then, the new mobile-enabled platform is more productive than other app-building tools. The idea is that many corporate apps can be implemented to run in the existing Salesforce 1 app, which perhaps more correctly should be called a client, while apps that need to be deployed more broadly, such as to consumers, can be built using the Mobile SDK and deployed to the App Store or Google Play.

Developers of course are used to these kinds of claims and will be sceptical. Still, if you have adopted Salesforce to the extent that all your users are on the system, then it might make sense to build apps with Salesforce 1 and have a lot done for you, including user management and authentication.

There is talk at Dreamforce of the “app gap”, the fact that typical enterprises currently have most of their apps designed for the desktop, but are planning for most of their apps to be mobile. That gap is an opportunity for Salesforce 1.

Against that, note that apps built with Salesforce 1 are not portable to other platform, and there are the usual questions about the extent to which businesses are willing to entrust their business to a third-party cloud platform, and if so, which cloud platform is the best choice.

Is Salesforce 1 the same old stuff repackaged, or something new? It is a bit of each.

As an aside, the focus here on iOS and Android will not be helpful to Microsoft/Nokia trying to sell Windows Phone in the enterprise. You can also understand why Microsoft is partnering with Xamarin to enable its .NET, C# libraries to work on iOS/Android. If enterprises are going mobile and largely not using Windows Phone to do so, Microsoft has no choice but to give full support to those rival mobile platforms.

amazon, cloud computing, software development

Catching up with Amazon’s cloud services

November 7, 2012 onlyconnect

I attended Amazon’s AWS (Amazon Web Services) Update in London. This was not a major news event; more a chance to catch up on what is new with Amazon’s cloud services, the dominant force in cloud computing infrastructure.

One thing that caught my interest is the speed which which Amazon is rolling out new features. The pattern seems to be that one or more significant features are rolled out each month. The session in London covered announcements since July 2012, with new stuff including:

DKIM signing for the Simple Email Service
High I/O EC2 (Elastic Compute Cloud) instances
Cross-origin resource sharing for S3 (Simple Storage Service), lets web apps interact directly with S3 content
Amazon Glacier service for archival storage
Binary data support in DynamoDB
SQL Server 2012 in RDS (Relational Database Service)
Provisioned IOPS (1,000 to 10,000 IOPS) storage for RDS
New instance types and price reductions – there are now seventeen types of VM, see the current range here.
General availability of Storage Gateway, which lets you attach cloud storage to your local network via iSCSI, with local caching for performance.
Ruby support in Elastic Beanstalk
Completely rewritten SDK for PHP using modern coding style
Consistent BatchGet for DynamoDB
Increased Provisioned IOPS for EBS (Elastic Block Storage) to a maximum of 2000 IOPS

What I want to highlight is not so much the features themselves as the pace of development, which is impressive.

There was considerable discussion of Provisioned IOPS which let you purchase fast data traffic between your application and your storage. This can have a dramatic impact. Netflix used it to reduce the instance count and eliminate Memcached caching from their application. Increasing performance is another route to scalability.

Reserved instances are interesting. If you reserve an instance for a period, rather than paying as you go, you save up to 63% but lose the benefit of down-sizing on demand. However Amazon has also created a marketplace where you can sell unused reserved instances. It is all smoke and mirrors for Amazon; a reserved instance is just a billing mechanism. It collects 12% of any resale though.

Elastic Beanstalk also got some attention. I have always thought of this primarily as an auto-scaling feature. However, the discussion focused more on ease of deployment. The two are related, since Elastic Beanstalk has to know how to automatically deploy your application in order to scale it automatically. It is “AWS for the lazy”, we were told.

Amazon is getting high demand for node.js on Elastic Beanstalk – not available yet but watch this space.

There was a session on CloudSearch which left me unexcited. This is in effect another type of cloud database designed for search with relevance ranking, field weighting and so on. However it is not trivial to implement; you will have to work out how to feed CloudSearch with data in its SDF format, matching what you want to search, and how to keep it up to date.

I would have liked to hear more about the DynamoDB NoSQL database manager which is proving a popular service.

If you want to track AWS as it evolves, I recommend following the official blog.

adobe, design, development, flash, professional, software development, Web, web authoring

Adobe results: 200,000 Creative Cloud subscribers and an impressive transition

September 20, 2012 onlyconnect

Adobe has released its quarterly figures for its third financial quarter 2012. The figures show the success of Creative Cloud, Adobe’s subscription-based model for purchasing the Creative Suite applications, including Photoshop, Illustrator, InDesign, Acrobat and Flash. Total revenue is fractionally up on the same period in 2011, from $1013.2M to $1080.6M.

Adobe reports over 200,000 paid subscribers and 8,000 new subscriptions per week, compared to its projections of only 5,000 per week.

The Creative Cloud model has several advantages for Adobe. First, it gives assurance of a steady continuing income rather than the pain of driving a 2 year upgrade cycle. Second, it forms a platform from which to sell other products and services.

Adobe also says that its publishing platform, the Digital Publishing Suite, now has 1,100 customers distributing on average 125,000 publications daily, mainly to the iPad, with over 40 million delivered to date. This is good business for Adobe since it generally charges a fee per download.

The slight downside for Adobe is that the launch of Creative Suite 6 delivered lower initial revenue than is usual for a new launch, because customers are transitioning to the subscription model. That is not really a downside, but rather a sign that the strategy is working.

What impresses me about Adobe is how well the company has survived the decline of Flash and the relative failure of its efforts in enterprise applications (the digital enterprise segment is now subsumed in the figures into “Digital Marketing”). The segment breakdown for the third quarter looks like this:

$millions

Digital Media (Creative Cloud) 769.1 (71%)
Digital Marketing (analytics etc) 257.1 (24%)
Print and Publishing 54.4 (5%)

Think back a couple of years. Adobe was dependent on sales of shrink-wrap software and had a range of products which pivoted around Flash as the universal runtime and rendering engine. Now it has some claim to being a cloud company – though of course the primary benefit of Creative Cloud is in desktop software applications that you download – and in place of Flash it it betting on HTML5, together with its ability to compile Flash-based content into native applications.

The transition is not so easy for developers who invested in the Flash platform, coding applications in Flex and ActionScript. Adobe has stopped developing Flash for mobile, even on Android and other mobile platforms where it is not blocked. Still, if that has pushed developers into targeting HTML5 earlier than they would otherwise have considered, it may not be a bad thing.

Tech Writing