Tuesday, September 23, 2008

On OpenSolaris Change

When I noted that Sun's plans for OpenSolaris threatened the Solaris ecosystem, I got a mixed bag of comments.

Some of the comments missed the point, which is that compatibility (across the board) is a key strength, and that producing something that forces what is essentially a new platform on the world will drive away old customers without necessarily attracting new ones.

The key point is compatibility. And while modernization is essential (I'll come back to that later), it is possible to do it compatibly, in an evolutionary manner, rather than doing a rip and replace job.

Evolutionary change allows you to keep existing customers who thus have an easier migration path; makes it easier for new adopters who can tap into the existing skills pool; and allows the improvements to be fed back to older (ie. current, such as Solaris 10) releases which still have a long service life ahead of them.

Replacing the packaging system and installer from scratch is just something you should never do. It's probably cost the Solaris/OpenSolaris ecosystem about 2 years, and we can only hope that we can eventually recover in the way that Firefox did after Netscape's mistake.

6 comments:

Kristofer Pettersson said...

If you trust all things you never should do you would end up doing nothing. Solaris was becoming irrelevant fast. Replacing the installer and turning the boat is risky indeed, and it takes a brave leader to do it. Stop whining about it and fix it instead. It will rock if we let it. :)

Rand said...

When it becomes obvious that a major change is necessary, the issue becomes "How can this change be implemented in such a way as to minimize disruption to those who use the current system?". Change is more tolerable to some groups of people than others. It is important to consider the needs of those people who might be seriously impacted by the change. Some kind of transitional system may needed to smooth things over. For example, a legacy interface to the new system for at least some commonly used functions. Of course, at this point we're speculating because we really don't know what is planned for the next major release of Solaris.

Anonymous said...

Compatibility will be there with Solaris Branded Zones. Solaris 11 will provide branded zones for solaris 10, 9, and 8. How much more compatibility could you want? There will even be branded zones for linux. That is SO much compatibility, out of the box. And that's not all. Solaris will have xVM for virtualizing windowns (and linux, if neccessary). So I can't see the validity of the statements you are making in these blog posts.
BTW, there is no progress without change.

Peter Tribble said...

kristofer - I'm not advocating doing nothing. Taking a couple of years to implement a brand new system is simply handing the opposition a couple of years lead; all this time and effort would have been better spent fixing what we have, which could have been finished ages ago and then had the additional weight of the existing installed base behind it

rand - the thing is, we could have done all this (and far more radical improvement) without breaking things. One thing Sun have been terrible at is taking something that isn't good enough and improving it. They would much rather throw things away and start over - history is littered with the wrecks of companies who've lost their market position while doing so.

Anonymous - branded zones don't help. Not only do they not meet our requirements (one key feature of Solaris 10 is zones; you can't run zones inside zones), it still requires the introduction of a whole different infrastructure which is the last thing that's going to be welcome when everyone's trying to simplify their infrastructure

milek said...

Peter - you won't have to do zones inside zones - what's wrong with several branded S10 zones running on S11?

Kimmo said...

The analogy to Netscape is flawed, we're not talking about recoding Solaris from scratch, we're talking about replacing patently outdated parts of it. As a multi-year Solaris admin I sincerely think patching and maintaining it is a massive PITA. It has been a PITA since I first started working with it many a year ago and it's a PITA today.

The installer and packaging needed change very badly and I for one am very optimistic about what the future will hold. My personal OpenSolaris use at home etc has been limited to my own server and I'm pretty stoked over how much less work it has been, and I have all those great features you mentioned in your previous post as well. Sure, there was a new learning curve, but that was by no means all that onerous.

I doubt if any Solaris shop out there is happy about maintaining 10 at this point. Oh, we can all do it (I hope), but I for one definitely look at my Ubuntu workstation with near envy considering the sheer ease with which I can maintain that by comparision.

A Solaris that is easy to install, maintain and also contains all the goodness of ZFS, native CIFS, Dtrace and all those other good things seems to me to be a far easier sell than one with an ancient anachronism for package management.

Oh and didn't the Netscape rewrite eventually become Firefox? ;-) It may or may not have been a bad decision for Netscape in the short term, but in the long term it was of quite the benefit for, umm, people who don't have the sense to run Opera... :-D I don't see Solaris failing or even stumbling very badly because bad parts of it get rewritten into something modern and usable. Especially so since we're still talking about only part of the complete product being ripped out and replaced.