 |
Tuesday November 16, 2004
Martin is from Venus, Sun is on Earth Update: After thinking about it, I modified the title of this blog entry from "HP" to "Martin". Martin was really speaking as himself on his own blog and not as a representative of HP.
This is a response to Martin's LinuxCIO blog. I'm tired and it is late, so this is not exactly a full-throttled response.
When Solaris becomes "open sourced", please remind me to revisit this topic. What I love about RSS feeds is they exist somewhere on the internet ... permanently.
Is it Really Solaris? Is Linux running on Itanium really Linux? [Of course it is.]. I'll admit that I don't tell my customers that Solaris binaries compiled for X86 hardware don't run on SPARC. I also don't tell them ice cream is cold. Stating the obvious is not the conversation our customers want to have with Sun. Solaris is a rock-solid industry-accepted operating environment that implements many industry standards. From Sun's view point, Solaris is Solaris is Solaris. We do not make the distinction between X86 or SPARC unless it is a required distinction in a conversation. Generally speaking, if a developer writes code to open standards, it will compile on SPARC or X86. From a customer's point of view, they have a choice of Solaris running on X86 or SPARC. All the features, minus hardware differences, exist on both.Is it really open?: I can't find anywhere where Sun has talked about open sourcing Solaris compiled running on X86 only. Let me answer this question bluntly, yes Solaris is equally open on both SPARC and X86. Why? Because ~95% of the code is the same (same base source code). Only code used to manage the hardware differences are different. Read yesterdays blog to read my brief comment on open source Solaris. We really want to do this (open sourcing) right, where many of our competitors simply do not want to do it at all.
Who can check in code?: Just wait and see. It depends on the license.
Why I don't think Sun will GPL: I do not know if Sun will GPL or not. We'll both have to wait and see.
Linux inside? Well, maybe: No, Solaris inside. Linux libraries and applications on top. And those Linux apps running on top of Solaris benefit from zones, resource management, dtrace, zfs, etc. I think we have roughly 600 certified applications Solaris on X86 with a 300% growth rate. I would love to compare that with Red Hat apps certified on Itanium. If a customer really wants to run Linux on v20's, we'll be happy to accomodate them. In fact, we can offer them the Java Enterprise System on Red Hat v20z's/v40z's and support the entire configuration.
Is Janus Indemnified?: Janus (Sun bits) is. We took a $92 Million dollar hit on behalf of Java users due to the Kodak patent case. We know how to indemnify customers who deploy our technology and we step up to the plate. We leave it up to Linux distributions (like Red Hat) to do the same for their customers.
Applications need developers/Where will they come from?: Open source != Linux. Let's break this down a bit. We are talking about CIOs here, right (LinuxCIO)? How many of those 12,000 applications are production deployable applications? How many will solve the problems that CIOs care about? Most of those CIO-important applications already run on Solaris (on X86 and SPARC). And for those that do not? Well, many (perhaps most) will compile on Solaris. Definitely the popular ones. I personally run MySQL and Apache on Solaris (X86 and SPARC) with PHP and GeekLog. Feel free to go to Sun Freeware or SunSource. We know all about open source.
What problem is Sun trying to solve?: How about customer problems! We have many customers that are very fond of Solaris and the innovations it contains, yet want the advantage of commodity X86 hardware at certain tiers of their architecture (for some customers, it may be all tiers).
Ah, what about TCO? Sun is not selling just Solaris support plans. Solaris does not live in a vacuum. Taking Solaris out of the context of a larger system means that you do not understand Sun and our business. Sun is a *systems* company. We do not want to divide and conquer. We want to unite and simplify. Put Solaris in the context of hardware, software, services and partners. Sun's Client Solutions Organization and our Software Elite partners help the customer integrate the system in to the customers environment. We don't think the only answer is to integrate 15 different non-integrated products from 15 different vendors on-site at a customer. Our goal is to integrate the system in the "factory" and then take that integrated "system" to the customer and use well-defined architectural patterns to integrate that "system" into the customer's environment. That gets us a long way to a solution. Add in our valuable ISV's and our partner's domain expertise and we finish the solution. No, you do not have to buy in to the entire system that Sun offers. Customers can use different "open/standard" products in part of the system, but I personally think that should be the exception, not the rule.
In the end: In the end Linux is not the only answer. It is a possible answer. As (perhaps) is HP-UX. As is AIX. As is FreeBSD. As is Solaris. Open sourcing Solaris means we can let our community get more involved. That is a good thing. In the end HP wants to throw stones at Solaris when they sell proprietary HP-UX and resell proprietary Microsoft Windows. Martin, perhaps some introspection would be beneficial. (2004-11-16 23:08:51.0) Permalink Comments [6]
Sucking info through a fire hose
I have been getting to know some of the Solaris kernel engineers a bit while I am getting hit with a fire hose of information. For those of you who think Sun is not committed to open sourcing Solaris, don't doubt our intent. The *last* thing Sun wants is to not do this right. After we bombarded them with questions, the team involved with open sourcing Solaris made things positively clear to me. No, I don't know the license :)
I also got a chance to get a deeper understanding of some Solaris features, particularly SMF, DTrace and Zones. Zones I have gotten proficient at a while ago, but I did pick up tidbits of good info. I may ignore my own advice and add some SMF functionality to the demo before I clean up the code. I already have the UI thought out. TBD.
Not much time for blogging. Back to class ...
(2004-11-16 06:30:11.0) Permalink Comments [2]
