port7 – co – uk

In the old days it was so easy, if you wanted to decide on what platform to host an application, you had the choice of Windows on x86 or AIX on Power (or some other UNIX on RISC), the differentiation was a in both the underlying hardware and the OS

Now as lots of companies are getting rid of the proprietary UNIX in favour of Linux. The rise of x86 is inevitable and well underway, a Gartner report from 2010 says;

Only one processor technology, x86, will grow in share beyond this decade as volume commodity markets continue their growth. However, the Unix market, with its RISC/Itanium chip technologies, will decline

Source: Impact of the New Generation of x86 on the Server Market

Now you can have either Linux or Windows on the same underlying hardware. How do you choose on which platform to run your application, as it will run on both. So do you run Oracle on Windows or Linux?

Which one will give you the better performance? Which one will be more scalable?

I think Linux has won the battle over proprietary UNIX but now the battle will soon revert to the old Windows vs Linux, and is much more stubble and with Windows

You could argue vertical scalability, but that’s not so easy as organisations try to simplify their infrastructure on small reusable units of compute power, and moving towards horizontally scalable applications.

Could it be functionality, and how ‘cloud ready’ the operating system is? And what makes a OS cloud ready?

It’s certainly a problem I am currently pondering, and one I think will increasingly be asked.

Recently I have been fortunate to do some architecture work for putting a Linux capability on System z, specifically z/VM.

One of the things that has struck me is what a truly strange beast this Linux on z really is.

First of all, if you look at the way regular linux is developed, there is a comminity around a number of upstream projects, that a distro like Fedora or Debian take and release. Then for enteprise market you have companies like Red Hat who come along take a Fedora release and polish that into RHEL.

Where is the Fedora for System z?

One the benefits of Linux that are sold to the enteprise is that there is this vendor agnostic community behind it all.

Then you have to think, what will you run on Linux z? I have an application say written by a company in something cross platform like J2EE, ask the vendor if they will be able to support it on Linux on z, I think there will be few who would.

The only stuff I have seen running on Linux on z, is either homegrown (and so home supported) or from IBM, who have a vested interest in selling the hardware. I know Linux on z offers great consolidation oppotunities, and the eco credentials are great, but what are you going to run on it???

Had a good day today, down at Red Hat’s UK offices in Farnborough. Nice offices and great location with the old airship hanger outside the window!

Reason I was there was for work where I am helping my current client to set up a Linux capability, and we are reviewing Red Hats capability to support Linux for a large financial organisation, which you would think is a no-brainer, but life is never that straight forward.

One of the beauties of Linux is that because it’s Free Software, you have the unique situation in that anyone can support the software. It’s always nice seeing senior IT industry veterans realise that.

The message from Red Hat was good, and I was glad to see them emphasise the community aspect of their business, and how important that is to not just them but their customers. Hopefully people may see the value of end users feeding back to the community, now that would be cool!

Recently been thinking, that if there is one thing that Open Source Community do really well is collaboration, and building community. Something that very few ‘Enterprise’ places do really well.

Considering why can’t an Enterprise do that, why can’t they? Is the question I am asking.

If I was starting a Open Source Project outside of work, I would be working the online community, frequenting IRC Channels, Mailing Lists, and Blogs of related projects. Announcing my new project and gathering a team around that.

Things like that just don’t seem to happen in Enterprise, sure there are no IRC Channels, or Mailing Lists. I guess the equivalent are teams, both support and project specific. To but in on those would be difficult.

Why can’t we do it?

Probably a number of you reading this will now have seen the announcement at Google I/O about the new product from the makers of Google Maps, what they are calling Google Wave.

What Google Wave is attempting to do its effectively re-invent E-mail, they asked themselves, what would E-Mail look like if it was invented today? The answer is something that looks like the lovechild of Gmail, Wiki, SubEthaEdit, Twitter, and Sharepoint. I think it could catch on.

As someone who makes his daily bread in the very conservative world of corporate IT, I can’t see it making inroads here (which is a shame), though the fact that Google are releasing the code and allowing people to host their own web-servers might make that interesting, especially with the federation aspect of the protocol.

I for one can’t wait to give it a test go, and I may even deploy it in some of my clients, if it grows in maturity.

Well done Google (again)

So yesterday I received a sales call, not a surprise as I get lots of them each day. But this time I was pleasantly surprised to discover that is was from Sirius IT, for those that don’t know Sirius are probably one of the more successful FOSS consultancy firms here in the UK. Having been around the Linux community since 1996 I have been aware of the good work they have done, and the growth they have enjoyed.

Unfortunately I couldn’t give them much work, as my current client is pretty much well served by a big outsourcing deal with Capgemini. Though I was pleased to tell the chap that they already use a significant amount of FOSS in the IT estate, and find it works very well thank you.

What encouraged me about this call was just that it indicated Sirius IT was growing into a serious contender in the UK IT industry. They have obviously started this cold-calling strategy to win business, and I hope it works from them. It certainly makes a change to hear a company calling and specifically mentioning Open Source software. In the current climate, I really think FOSS has a lot to offer business in the UK, and hopefully initiatives like this from Sirius will do a lot to make that happen.

So well done Sirius IT, I hope to send some business your way in the future, maybe not with the current client but maybe another one.

So having a number of chats with various people about what we would like to see in a corporate twit platform.

Following from a blog post by friend and colleague Mr K, I have been thinking, could we build something?

That naturally progresses to, what do we want to build, how would we build it, and with what would we build it?

Addressing the first question, could we build it? The answer is of course yes, it’s just software and assuming we have a sensible set of requirements and some clever bod’s anything is possible.

The second question is what do we want to build, is possibly the tricky one. Why not just use Twiter? Well essentially my problem with that is that it is a hosted service, and for any corporate having potentially private and sensitive company info hosted on a server outside it’s control is just simply a no-go. Also on the privacy aspect, your company tweets could appear in the public tweet list.

So what do we want?:

• Ability to ‘follow people’
• Ability to create groups, and also to follow groups
• Ability to search posts, both all and people you are following
• Options for backend auth (LDAP, AD, OpenID)
• Logging, so all posts can be logged
• Ability to post and read by various means, IM, Web, SMS, etc
• Scalability, this should be able to scale to corporations of 100,000+ users

So with with what should we build it, given the requirements above, could this be done with the LAMP stack? Or do we need a proper daemon written in C? Not being a developer, I don’t know these things. My guess is that LAMP could do all the above, though the scalability requirement might be tricky (or not!)

Anyway I am throwing these thoughts out to the ether, feel free to comment, volunteer to do it, or just shoot me down

Reading an interesting post about community equity, some interesting thoughts there. I have been wondering myself recently about how business and community can help each other, getting involved in communities I think is a big step forward.

I may post more about this in the future….

Dear Lazyweb,

Can someone please tell me what the difference between NLPOS9 and IRES2 is? And what do I get over NLPOS9 by going with IRES2?

Is it drivers, is it admin tools, is it support agreements? What is it!!??

Is anyone aware of any Debian derivatives that are targeted to the Retail market? Looking at the current market, it seem that Novell has it cornered. That can’t be a good thing!