Archive for the ‘Linux’ Category
We released a new version of Landscape our management service for Ubuntu last week. There’s a slew of new features including server templates, simple upgrades and enterprise authentication support. Whether you’re managing a few systems or as many as Google the new features make system administration simpler!
Landscape’s objective is to make managing and monitoring hundreds of Ubuntu systems as easy as looking after one. Whether you’re managing some Ubuntu desktops, or looking after a Web server farm Landscape lowers the complexity of administering those systems: no-one wants to apply patches to hundreds of machines manually! For IT managers this means that Landscape makes system administrators more effective and efficient. Landscape also ensures that deployed Ubuntu systems are secure with maintenance patches and upgrades.
Landscape is provided as a software service so every six months Canonical releases a new version that is available to all subscribers. There’s also an on-site version available to customers that have security policies or regulations that prevent them using a SaaS management platform. In line with Ubuntu 10.04 the main features of the new version are:
Many sites have sets of servers that do similar jobs, for example “web serving“. Ideally you want those machines to have the same set-up reducing management overhead.
The ability to create templates of the packages installed on a particular system and then apply those to different machines makes it easy to replicate a standard install. It also ensures that you maintain consistent profiles across your systems as time goes by. Finally, if you need to re-provision or expand resources you can use profiles to ensure it’s a repeatable process. Package Profiles is really great for managing configurations.
If you’re managing more that a handful of Ubuntu systems then doing upgrades is going to take a lot of time. Whether that’s every six months in time with the standard releases, or every two years for the LTS releases, it’s a significant commitment. To reduce that overhead you can now do upgrades between releases using Landscape.
Upgrades between releases are always complex so this doesn’t remove the need for backups and careful attention. Nonetheless, if you’ve used Package Profiles, it will be easier to test an upgrade on a test system and then when you’re happy apply it to all the deployed systems using the same package profile. Rather than having to access every machine and do the process by hand you can upgrade a group at a time.
Enterprises commonly have a corporate standard for authentication such as LDAP or on a Microsoft Windows network Active Directory. The new version of LDS connects to these systems authenticating administrators from the existing authentication system. This ensures that customers can simplify their authentication set-up and enforce authorisation from a single corporate directory.
What software will Canonical provide support for? That’s probably one of the questions you were asking if you read my previous post about commercial service subscriptions and bug resolution. Or perhaps not, but it’s a rhetorical device that suits me for this post!
Generally speaking for an application to be supported as part of a service subscription it has to be within the Main repository. This is because applications within the Main repository receive public maintenance (bug fixes and security updates) for the life-cycle of the release.
In order for an application to move into Main it goes through a stringent security and quality assurance assessment. As part of this review Canonical’s engineers inspect the code and ensure that they are able to maintain it. Consequently, those engineers also provide bug-fixes and maintenance for Canonical customers.
I find it interesting that generally the ability to maintain and fix code is one type of developer skill-set, while writing new features is a different one. Colin Watson recently told me that an early manager had told him that there are two types of developers in the world, those that create things and those that finish them off. Intuitively that feels right to me and by definition a distribution is focused on the latter where integration, polish and quality assurance rule.
The second issue is how do you know which software is covered within the Ubuntu service that you subscribed to? Some Linux distributions deal with this by covering all the software that they physically ship to customers. However, in Ubuntu’s case most users receive the software electronically so this doesn’t work. Second, the Main archive and seeds are relatively fixed and don’t map well to a subscription service for a particular target market. Essentially this means it’s hard to reflect the services within the technology.
Consequently, when a customer purchases a particular service subscription they receive a Service Description. This describes the scope of support, the bug-fixing coverage, the legal indemnification, the software components covered and the response levels. For example, a consumer desktop service wouldn’t cover complex integration problems with a Microsoft Windows network, while this would be critical for a corporate subscription designed for customers with legacy networks. Effectively, the description tries to describe the types of use-cases and categories that are covered.
I hope this has given a bit of insight into how Canonical does support and bug-fixes for our customers.
If you have a commercial subscription service for Ubuntu how do we prioritise fixing bugs? This was at the heart of a conversation I had with a customer recently.
For business users Ubuntu’s advantage is often flexibility. Adding another system to the data centre is simply a matter of starting it up. This contrasts with proprietary UNIX and the other commercial Linux vendors where license management creates deployment friction.
Nonetheless, it’s hardly “free” if you can’t use the software. And Ubuntu, like all software, has bugs and issues – particularly when you’re using it in a complex environment. To resolve these issues professional users need access to expertise when there’s an issue.
In the proprietary world the license agreement commonly includes support so the customer presents the bug and they should get a resolution.
Ubuntu’s free nature presents a more nuanced picture. Every Ubuntu user is able (and encouraged) to put bugs into Launchpad. Many of these bugs will be resolved by Ubuntu community developers or Canonical’s developers as we work on the next release of Ubuntu.
Nonetheless, any individual bug is a needle in a haystack. Ubuntu receives vast numbers of bugs from our user-base so there’s no guarantee that any individual bug will get a response or a resolution. There’s inherently no prioritisation of one user over another as all members of the Open Source community are equal. Additionally, bugs are generally resolved in the version of Ubuntu under development rather than the one that the problem is reported against. The need for certainty of response and resolution is the value of a formal relationship with Canonical.
A service agreement means that the customers bugs are guaranteed a response, that the issue will be dealt with by an Ubuntu expert and that the issue will be prioritised. For Canonical engineers customer bugs are prioritised over general development work and are split into categories by urgency.
Initially when the customer presents the case the GSS (Global Support & Services) team triage it and where possible come up with an immediate workaround. If the bug requires code development then it is escalated to the appropriate engineering group. This is where a resolution for the version of Ubuntu that the customer is using is created. This is generally delivered to the customer as a custom package for them to use immediately. The resolution is then integrated into the version of Ubuntu under development so that there won’t be a regression when the customer upgrades to the next release.
So flexibility is the Ubuntu advantage, and the advantage of working with Canonical is there’s a canonical resource for Ubuntu expertise.
This week I’ve been in China talking about Ubuntu 10.04 and learning how Ubuntu is used. I’ve learned alot and it’s been a great opportunity to encourage businesses and partners to extend their use of Open Source (Ubuntu in particular!). I’ve really enjoyed this trip and it’s been fantastic to meet so many new people.
Like everyone else I’ve been struck by the energy in China. Everything from the traffic to the rate of technology change is done at full pace. China is a country that is growing up rapidly, across the board there’s great confidence and a total focus on how to grow both in domestic markets and in global trade.
It’s a country with significant divides between the urban and rural communities. One project we’ve been working on is to provide a PC that can be used in rural communities and education. The objective is to provide a platform for those groups to improve their lives, careers and education. The team has done a fantastic job and it’s great to see Ubuntu being part of this project.
I’ve come away convinced that Open Source is an ideal solution for many of the different challenges. Of course it’s cheaper than proprietary solutions and that has significant impact. But, more importantly Open Source and Ubuntu can deliver tailored solutions from the desktop through to the server and into the Cloud. This ability to fit into many niches as a flexible solution and help drive innovation is a key advantage for Ubuntu users.
I’d like to thank Richard, ZengPeng and Juergen for showing me around and Fanny Yeh for helping me navigate the region (particularly with the volcanic clouds!)
Wavesat is using the Bazaar version control system for commercial development making it simpler and easier for their teams to collaborate around the world. It’s a great example of Open Source delivering cost savings and innovation to business users. We’ve recently put up a case study that gives more details.
Bazaar (Bzr) is a distributed version control system. It’s an essential tool for developers: there’s a great guide to revision control on betterexplained.com. When people state that there’s no innovation in Open Source, distributed revision control is one of the examples that counters this.
Bazaar is particularly well suited to distributed development because the concept is built-in right from the start. Perhaps it’s testament to the open source development process which is by its nature distributed. For a business like Wavesat that has developers based in different locations this means they can be more efficient.
Canonical sponsors the development of Bazaar because distributed revision control is critical in Open Source development. But, it’s also something that companies can benefit from so we provide commercial services for Bazaar. This consists of helping organisations migrate, along with providing support and training. For organisations with an existing version control system such as CVS or Perforce we help with the migration to a new work-flow using Bazaar on Linux (Ubuntu, RHEL, SLES) or a legacy operating system such as Microsoft Windows or Apple Mac. Check out the case study for more information.
As I write this I’m rocketing across Belgium shortly to transit under the channel on the Eurostar train. I’m crossing a continent rich in history and conflict. But to be honest, it’s an antiseptic journey. Unfortunately, my laptop isn’t on-line, and without a connection it won’t relieve the boredom. Luckily my phone is online. It’s my phone that I’m using to stay connected to the Internet, making sure I’m not missing anything important and keeping me amused during the journey. With it I’m reading e-mail and RSS along with communicating through instant messaging.
As we move towards a world that is mobile and connected the phone is much closer to the type of computing experience I want than the traditional PC. Perhaps without realising it phones are becoming a very significant computing platform. We’re using an increasing number of them – billions of phones ship a year, millions of computers by contrast. And as you know emerging areas of the world are very big on phones! Finally, smart phones are becoming very powerful, Moores law and all that.
So if you talk about pervasive computing, the incarnation we have of that today is the smart phone: it performs some set of computing tasks and it’s always available to me since I always carry it. Consequently, if you’re interested in what the future of computing looks like then it’s going to have a lot of the characteristics of the phone.
I recently got a Nokia n900 which has a lot of the characteristics that I think are important in a pervasive computing platform. First, for all that we love touch you need data entry and that means a keyboard of some form. Second, if it’s pervasive then it better be well connected, because I move around a lot. If it’s an always-available computer then it has to be able to do more than one thing at a time, because I’m doing more than one thing at a time. Clearly, it has to have a wide range of applications and that includes full Web capability since lots of “applications” I access like Facebook are on the Web. And it has to be configurable, it’s my device and while it may be managed, I want to customise it.
So having set out my strawman it’s easy for me to conclude that the n900 ticks all those boxes. And it does, the N900 really is a good personal computing platform, it has a lot of the elements I listed. On the other hand, it’s not a very good phone!
OK, so I set the punch line up. Seriously though, if you just want a phone then you’re missing the point of the n900. Yes, the voice quality is good, and the integration with contacts and conversations is very nice, the embedded usage of Skype and ability to use VOIP/Skype is very interesting. But it has quirks as a pure phone, and anyway it’s noticeably large, heavy and the battery life borders on terrible.
As a mobile computing platform though it’s very cool. The more I use n900 the more impressed I am with the software package. Maemo feels very well put together to me, it’s fast, looks good and the UI has some interesting innovations. Maemo is Open Source and Nokia have been working to develop a community around it for a few years. It’s still raw in parts but if it keeps on maturing and benefits from Open Source innovation then it’s going to be great.
If you’re in the market for a new smart phone and you want something a little different, or you’re looking for something a bit more powerful and hackable than the other options out there then check out the N900.
Want to know Canonical’s secret business plan? Or find out the latest features we’re working on in Ubuntu or UbuntuOne? Then hop over to the Canonical Voices site. It’s a blog aggregator that provides a single location for Canonical employees to blog and engage with the wider world.
Many Canonical employees develop Ubuntu directly making them members of the Ubuntu community so their views already appear on Ubuntu Planet. However, there are lots of Canonical employees who work in other areas, such as with OEM’s, or on UbuntuOne, in marketing or with business customers. Canonical voices brings together everyone in the company and provides a single place where you can see the breadth of their views, opinions and thoughts.
As an Open Source technology company we’re working within a variety of communities; sometimes that means an Open Source project, but it could mean a group of users or a set of companies. So it’s important for us to be transparent and to engage in a conversation – encouraging understanding and perhaps sparking interesting ideas. Canonical Voices provides a space for that.
A connected point is that Canonical hires a lot of intelligent, opinionated and interesting people who are great communicators. Hopefully, Voices will provide a focus and context for those that want to blog, sparking everyone within the company to feel they are part of an organisation wide conversation. Personally, I’ve been reading Voices regularly for the last few weeks and I’ve already learnt lots of interesting things about other projects within Canonical.
I can’t promise that I’ll be any better at blogging regularly, I’ve already broken quite a few promises and resolutions on that front! Nonetheless, I’ve started aggregating posts about Ubuntu, Linux and Canonical over to the Voices site. Please check it out and become part of the conversation!
One of the most common requests the Ubuntu community asks for is a home server or small business server. This Beepstar post, The trouble with Ubuntu Server for beginners, encapsulates the argument nicely when the article says:
“95% of the would-be “nixers” are completely stunned, at that point when the Ubuntu Server installation states that it has finished and all that’s offered to the user is a black screen and a prompt line. Users … basically scrap the whole thing, install Windows and use … solutions which lack raw power but come with an comprehensive interface”
It’s certainly an interesting point, we can surmise that one of the things that heavily assisted the growth of Windows on the server was the Graphical User Interface (GUI) that came with NT 3.5 and NT 4. At the time the competitive product was Netware which was the dominant technology for providing servers in LAN’s, and networks were themselves reasonably new for small business networks. Windows rode the networking trend really well, and gave advanced technical users (rather than professional IT staff) the idea that they could run their own servers.
I’ll come back to the question of whether Ubuntu server should be trying to focus in this area for a moment, and just focus on technology problems we face in providing a home server. There’s two elements:
a. A set of common services
The use-cases are relatively straight-forward but the key is the integration. So we’d want thinks like basic file and print, with network services.
b. A nice user experience
An easy to use interface that can guide the user through the initial installation, but also the reconfiguration and management of the services.
We’ve been working on common services in Ubuntu server and ensuring that they’re well integrated and easy to set-up as this makes every system administrators life easier. So making LAMP easy to install, integrating the experience of attaching to a Windows Network and the recent e-mail stack work all make setting up common services easy and quick.
To provide a graphical user experience there are a range of options. There’s some well-known free software options, the two most well-know are E-Box and Webmin. There’s also commercial control panels such as Plesk which is used a lot by hosting providers.
It’s difficult to see a way to integrate one of these panels as the default way of adminstering an Ubuntu server as the impact on professional users would be dramatic. For various reasons these tools assume that you only manage the system through the GUI. So there’s no way to integrate them that would maintain the freedom of professional system administrators to manage the system using the command line interface.
Meanwhile, professional system administrators face a different set of problems. The shift of delivering everything through a web server and the introduction of virtualization and cloud computing is causing an explosion of server instances. So for these use cases the focus is on a small, efficient server with centralised configuration management capability.
The compromise may well be a small, powerful server platform aimed for cloud computing. Then a range of appliances (virtual or otherwise) built to meet the specific needs of both professional and personal (ie home) users. There’s been a few different community efforts along these lines and I hope we’ll see more.
The Eclipse team released a survey that shows Linux is the most popular deployment platform, outstripping Microsoft Windows. Ubuntu is the second most popular Linux deployment platform, just behind Red Hat Enterprise Linux (RHEL). On the development side the Linux desktop shows strong growth, with 27% of developers using it.
The survey itself had over 1,300 participants from a range of different organisations, you can get the full results from their site. Eclipse is popular with Java developers or developers not strongly tied to Microsoft technologies, so the survey gives good insight into their behaviour.
From the survey Linux is used by 27% of developers as their desktop environment, up from 20% in 2007. This is very strong growth, and shows that Linux continues to expand its reach on the desktop. Ubuntu is the lead distribution with 14.5% of users, RHEL/Fedora is next on 4.7% and then SUSE/Novell on 3.2%. Windows is still the largest platform but has lost 10% market share; it was 74% in 2007 and has now dropped to 64% in the 2009 survey.
The survey also asked what platform was used for deployment. There’s a dramatic shift from previous surveys as Linux now represents the majority on 43% and Microsoft Windows 41%. The breakdown is also interesting. RHEL/Fedora is the lead distribution with 13.1%, Ubuntu comes second with 12.0% and SUSE/Novell is used by 5.3% for deployments.
The desktop Linux trend is really good news as it’s another strong data point showing that Linux is growing on the desktop. Developers are power users who tweak and play with their environment, so if they’re satisfied with Ubuntu then it bodes well for other broader users.
For Ubuntu the deployment platform data is also good news as it shows continuing growth for the server. By it’s nature the server isn’t as flashy as the desktop, it’s strengths and capabilities are more subtle. It’s great to see that developers are appreciating that the server edition is capable of delivering the full range of enterprise workloads, including Java application serving. A lot of work has been put into making LAMP and Tomcat easy to install and maintain, to my mind this survey demonstrates that the work is paying off – we’re helping users deploy Java more simply.
Ubuntu is widely used as a server platform, across organisations of all sizes, particularly for web and application delivery. Those are the results from a survey of 7000 Ubuntu users that we released a while ago, in conjunction with RedMonk. The press coverage was good, and now it’s died down I thought I’d add a few other thoughts.
The first question has to be why we undertook a survey. Understanding how platform’s used, what users value and what features they want is incredibly important. We use that input to guide our development, inform our decisions and focus our attention. Open source development provides significant detailed and direct feedback, but surveys offer some other insights.
First, they’re a route to reach people that don’t take part in the development process directly – which is the majority of users. Second, the volume of input gives us a strong statistical basis for examining trends – 7000 users is a significant pool of input. Finally, it’s an objective confirmation to qualitative feedback that we receive through other mechanisms, this helps confirm assessments we’ve made or trends we’ve observed. If you took the survey thanks very much for your time and input!
What are the main take-aways from the survey? Well clearly our users depend on Ubuntu as a server platform, it’s used for serious production deployments. It’s used across a wide range of organisation sizes and industries. That’s important because it tells us that our efforts to develop the platform, and spread the message that Ubuntu has a strong server capability are meeting with success. It’s also a data driven stake through the heart of the myth that Ubuntu server is only used by small business or hobbyists.
Second, that our users value the key qualities that we provide – the heritage from Debian, the regular releases, the focus on a tight and efficient platform for all the common workloads. I was particularly struck by the range of workloads that Ubuntu is being used for. We see Ubuntu being used for edge of network infrastucture (DNS/Web), but also internal application delivery (Java app serving/Databases) and it’s starting to make its way into line of business (CRM/ERP) delivery. These aspects are important because they guide our focus for future development as we make our way towards the next LTS release in 2010.
Finally, the range of geographies and individual situations where the server is deployed is incredibly exciting. Between the geographical data from Shipit and this survey we can conclude that Ubuntu is used globally. From a large media delivery platform in Europe, to a school in the Phillipines, each is spreading Ubuntu and benefiting from it in their own unique way. In the long-run this reach broadens the tent of Ubuntu (and Linux) supporters and developers. If the next 8 million users come from the developing world image what we can accomplish!
My thanks to Nick Barcet and the server community who put the survey together – it was a lot of work but has been really beneficial. I hope we’ll continue to run surveys in the future so that we can build up a picture of how things change. If you’d like to see the results with a deeper analysis then pop over to Gerry Carr’s blog post to download the whitepaper.