It was close to lunchtime when my iphone buzzed with the SMS: “Want some FUD?” I had to laugh; while my teenagers are specialists in the new lingo – this errant 'Fear, Uncertainty, Doubt' message from a co-worker was actually an abbreviation for "food". But seeing FUD on my phone's screen reminded me of the months before Y2K – I was working in IT for a telco, and we were feverishly updating all our server equipment to ensure we wouldn't run into the dreaded short date format issues.

Scroll forward 9 years. Here we are, and IT shops are looking at their aging server and storage inventory – many acquired in '99 with Y2K budgets, many facing end-of-service-life, many not meeting current or projected performance demands, costing too much for power and cooling and taking up too much datacenter floorspace.

With the efficiency and consolidation options available today, it's easy to make the case that it's cheaper to move to a new server than stay on the old. So why does anyone hesitate in moving from their older systems? FUD – think of all the issues with moving to something new: painful learning curve, disruption, customized software, ISV apps. Will moving cause costly interruptions to business?

Sun offers two solutions to take the FUD out of datacenter upgrades:

1. Solaris 8 and 9 Containers;
2. Sun Professional Services.

Solaris 8 and 9 Containers are virtual environments for hosting Solaris 8 and 9 applications on a Solaris 10 box. They provide a Solaris 8 and 9 runtime environment with all the performance and quality improvements of the Solaris 10 OS (DTrace, ZFS, Solaris Resource Manager). Now you can upgrade hardware in one stage and your applications in another. Less pain, more time to plan. Containers are a "transition tool" to help port applications to Solaris 10 in comfortable stages (watch this great video with the great Joost Pronk in which he explains Solaris Containers).

And to go with our Containers we have our experts - Sun Professional Services. Our migration team analyzes your original Solaris 8 and Solaris 9 environments, creates a migration plan, and implements and tests solutions as stand-alone projects. Professional Services can easily test, implement and optimize future system and network architectures for our customers (like Barmer Ersatzkasse), while protecting their prior qualification efforts.

No worries. Sun, we take the FUD out of migration. Now if I could just get some lunch.

Today at Sun we're all bouncing off the walls because today Sun launches the Sun Storage 7000 Unified Storage Series (code name "Amber Road"), the world's first open storage appliances. Words like "disruptive", "revolutionary", "transformative" and "radical" have been used to describe the new Sun Storage 7110, 7210, and 7410 Unified Storage Systems. Deserved or hype? I can think of three things off the bat that argue for deserved:

• an Open Architecture
• ZFS Hybrid Storage Pools and
• unprecedented, unheard of Storage Analytics.

ZFS Hybrid Storage Pools are storage stacks made from a mix of DRAM, Flash/SSD and SATA. ZFS manages this storage hierarchy as one transparent pool optimizing it to leverage the best attributes of each device. This optimization means the best performance (at about 25% the cost of traditional storage) and best energy-efficiency possible. ZFS's optimizations yields a 3.2 times faster Read IOPS, 11% faster Write IOPS and a 2 times faster raw capacity. ZFS not only optimizes for speed it also constantly runs data integrity checks to prevent any data corruption. It's not only fast, it's good.

Storage Analytics The 7000 Class Systems has a browser user interface (BUI) that radically simplifies administration tasks like configuration, maintenance (including hardware), checking shares (the 7000 line exports files systems as shares) and status (current usage of CPU, memory, storage, network, services, hardware, CIFS, NDMP, NFSv3 and v4, and iSCSI - it's pretty comprehensive and all on one page!) and, most wonderfully, DTrace analytics. In the storage world robust analytics on workloads in production just haven't existed. Now an administrator is able to look at a problem in real time - all while systems continue running in production. The Storage Analytics uses a drill-down analysis - checking the higher level statistics first and then going into finer detail based on previous findings. So, for example, things are moving along smoothly and suddenly performance is bad. With the Storage Analytics you can now ask: How many IOPS is the system doing? Which clients are causing a spike in IOPS? Let's say it's a CIFS protocol causing the problem; from that data point you'll then drill down and ask, Which Windows Client is going crazy? Is it doing more reads or writes? Which file is it reading or writing to? Before you would have been stopped at the second question. Now life is good. An administrator can quickly identify and diagnose system performance issues, and debug storage and network problems. Find it quick and fix it quick without shutting anything down. Pretty amazing. So far ahead of anything else available, you might even call it disruptive.

Sun doesn't stop at great open architecture, open storage appliances, revolutionary features like ZFS Hybrid Storage Pools, and get-it-no-where-else Storage Analytics. Sun follows up the 7000 class systems with great services. Our Professional Services is ready to help your storage migration with our Sun Unified Storage Data Migration Service. Sun's experts will migrate your storage systems quickly and securely saving you time and bringing you the full benefits of all the 7000 series features.

Most of my previous blogs have been about our Storage strategy at Sun. This past summer I moved into a similar role in Sun's Software group. What does that mean? Well even though I've got a new job, a new title, a new boss, and a new BU, it all feels pretty familiar.

While storage is all about data...keeping it, replicating it, archiving it, retrieving it, it's the software that provides the special sauce in managing data efficiently. Which is why when we talk about data integrity, scalability, and performance for unstructured data...ZFS is the answer. ZFS ROX. It must be true. There's even a license plate to prove it.

But what about structured data....how 'bout a database? How 'bout the world's most popular open source database? Last week, Sun announced we are acquiring MySQL. Rich Green blogged about how this is a good match on a company level - we're both big believers in open source, both have active contributing communities, both focused on the web economy. That's all good. Again, big changes, new stuff, but also somehow familiar.

But how do we match up on software? A good fit? ZFS and MySQL databases are a lot alike. Twins separated at birth? Well, customers use the same adjectives to describe the two: easy to use, reliable, flexible/scalable, great performance. Nice. Happily the differences between the software, and they are different, not twins but rather siblings who actually get along (does that happen?), complement each other: ZFS managing the storage of unstructured data and MySQL managing the use of structured data.

Kinda like my two teenagers - Danielle the structured just about taking over her college; Tara the unstructured defining her own home schooling plan... Absolutely love 'em both!

We've discovered the same thing about storage systems. A few years ago you could get storage in just a couple flavors - the original block devices or file storage that came on the scene in the 1990s. Both had just a few data services floating around the bottom like olives for additional flavor. But now that the world is starting to build storage systems using standard server technology and general purpose operating systems, more and more flavors of storage are seeing the light of day. Flavors dictated by the problem facing the customer, not by the layout of data on some physical media. Flavors formed by the application dealing with the customer problem, not by speeds and feeds.

This week a team of specialists in the UK tackled a big problem faced by our European telco customers - the controversial EU Data Retention Directive with a new Secure Data Retrieval Server that leverages an X4500 server, Solaris 10 operating system, and some partner software build purposely to solve this type of problem. The UK team didn't have to spend years negotiating a product spec with large engineering teams - our engineering teams and our partner community had already empowered them with the ingredients to build a compliance solution specific to the needs of those European businesses dealing with the new EU legislation.

Someone asked me today if the announcement of the Secure Data Retrieval Server means we're no longer committed to Honeycomb. Nothing could be further from the truth. Fact is, our coffers are full of ingredients, and while our Honeycomb team is working hard with customers and ISVs in healthcare and digital libraries, our team closest to the problems facing European telcos found the combination of the X4500, Solaris, and CopperEye created a great solution.

Because when a vendor has the right ingredients, customers have choice.

Besides the technology, one of the interesting things in the storage industry right now is the general consensus that data management needs to be handled by an end-to-end system, not just a storage box. Take virtualization - instead of virtualizing arrays behind a block storage controller, customers might choose to automate storage processes to support server virtualization. And CDP - as Nigel says "A year ago you could swing a cat and hit a CDP product", and now people are approaching CDP more cautiously because they just don't understand their data well enough to know where they need those types of RPOs. BTW, Nigel is a cat person, so no need to call out the SPCA as I don't think he would actually swing one around by the tail. And then there's de-duplication - are you really ready to let a computer decide which data should be deleted? We all still have lots of work to do with our customers before automatic deletion of data - in order to save storage space - will be a reality.

And as my luck would have it, Nigel - a die-hard Yankees fan - came to Boston the one time this year when the Sox are losing and the Yankees winning. He tried hard to rub it in, but then again, the Sox are still over 10 games ahead of those pinheads, oops I mean pinstripes in the AL East. Ha! But Nigel did make an impression on the nice guy who drove us from meeting to meeting: Kevin's parting words were "Finally after all these years of driving, I've met a Yankees fan I like."

We've certainly tried that approach at Sun; we've even acquired companies to help us in those attempts. But we always come back to our core of openness, partners, and innovation. So while the traditional focus has been around creating the mother-of-all-storage-management-dashboards, we've been focused on inventing better data management engines, advocating storage standards and delivering application-enabled device management.

Now don't get me wrong, some customers still need that dashboard. So for years now we've offered one that is considered best-of-breed on that market. And we're happy to look to one of our strategic partners for the development of that product. And some in the traditional storage industry also don't give you credit when you partner for a piece of your portfolio - but partnerships are an essential part of our strategy at Sun.

Let's talk about our areas of focus:

1. Innovation in data management - with Solaris and ZFS, we've got the automatic transmission for storage. No more deciding how to lay out volumes, which RAID to configure, etc. It's all built right in, and it's all automated.

2. Storage standards - After spending decades in the networking world, where you're nothing if you aren't open, when I started working in storage a couple years ago I was amazed at the API swapping that was going on. Thank goodness for SNIA and the Storage Management Initiative (SMI). Finally vendors are working to comply to open standards, rather than hoarding their interfaces for themselves and special friends. At SNW last month, I participated in a SMI Executive Luncheon where we discussed not only where SMI should go next, but also our commitment as key storage vendors to that standard. There were 30 of us in that room, and the majority of us knew that our future success, and that of our customers, depended upon open standards. Sure you could still see some feet dragging, but the storage world is finally becoming open. At Sun we're also working really hard in the object and file spaces too - in the near future look for XAM and NFS changes that will make data management just that much easier.

3. Application-enabled device management - we've got lots of storage devices in our portfolio, and we're focused on ensuring they have the best device management possible. So there are two things going on here - first we're using standards to talk to the devices, second we're setting up the devices so that the apps using them will work well. On the disk side of the house, we have our Common Array Manager, which speaks SMI-S on the bottom to configure and manage arrays, and supports application profiles on the top. On the tape side of the world, we just released ACSLS v7.2 today (that's our Automated Cartridge System Library Software), which now uses PostgreSQL as its underlying database management system. And we're working to continue integrating all the great backup management software that came to us through STK (and it came to STK through their acquisition of Storability) with ACSLS - bringing the best of tape virtualization management together with backup management. Not to mention all sorts of data protection software and automated archive software, etc, etc. All that on top of a wide system of partners and an open source community.

Considering this is my longest blog entry yet, it's amazing there are those who don't think Sun has much in the way of storage management software. Perhaps we're just counting differently in the new world.

It was all sweet at JavaOne this week... JavaFX, OpenJDK , Java SE , Java Devices, Java games, Java shirts, Java everything. But who would have expected a Java storage system, with built-in search and a Java API? A Java storage system with an open source community?

This is a first for us at JavaOne. Our Honeycomb fixed content system has Java APIs (C too), so that an application developer can easily structure how they want to store and search massive amounts of data. And we're using OpenSolaris and other open source communities to help further define these APIs. We also gave out the Honeycomb SDK on these cute memory sticks at JavaOne.

Our developers are busy working to release source code for the initial Honeycomb client and server implementation. We're targeting Fall 2007 for that, so there's lots going on and lots more to come. Meanwhile, join our Honeycomb development team in the OpenSolaris community for some good discussions about where we should be taking the Java API for this project.

Let's set the background: First, we actually did grow up with respect to our storage portfolio (certainly a lot had to do w/ Sun's acquisition of STK). Check out the storage products we announced at SNW: our Low Cost Array, the library partitioning (ha – the only library on the market that can be shared between mainframes and open systems), our incredible file browser that lets you see across tiers of storage. New conversations are always easier when people believe you understand the old conversations. Second, the stage continues to change with exploding data growth (exabytes, exabytes, exabytes) and extensive retention periods (decades, decades, decades). When it comes to data, it's just not the same world it was last year, or the year before, or the year before that... And last but certainly not least, general purpose technology has caught up with the performance and reliability demands of storage - there's little need to build the world's storage from scratch anymore. Out with the custom ASICs and boards, out with the real-time embedded kernels. In with x64 servers, volume CPUs, Ethernet, and the general purpose operating system. So if you can build storage from general purpose hardware and software (check out the Thumper), what's next?

Open source, of course. Once you move from the proprietary to the general purpose, you can take advantage of all that world has to offer. Which is why on April 10th we announced an open source community for storage – because we're making the entire storage I/O stack within Solaris available via the opensolaris.org storage community. So customers and partners working to solve the world's toughest data management problems now have access to the source code and the developers of the world's best operating system.

Now that we've hatched our storage open source community, interest from storage analysts, press, customers, and partners is overwhelming. The ugly duckling has truly become the swan.

And talk about using the proverbial can opener on storage. We're opening up storage today with our new open source community for storage developers. You'll find filesystems, data services, drivers - all in an open source operating system that scales, is secure, highly available, and reliable.

Too much more to talk about now. I just heard those Opening Day fighter jets fly over our Burlington campus. While I'm checking out our opensolaris storage community, I gotta tune into MLB.com for updates on the Red Sox. BTW, ever check out how MLB.com actually feeds us all that real-time data?

Earlier this week I was with an account team discussing Sun's whole portfolio with a global bank. First the server guys pitched... I love listening to SEs talk with customers - each one has a unique and interesting twist on their area, with cool tools, interesting data [not to mention insightful opinions that we can wrap back into product strategy] and excellent technical explanations that get to the meat of a customer's decision. Oh, the delicate balance between chip frequency and memory speed leading into CMT was a delightful discussion! Then the software dudes took over - Java, Solaris, open source, more Java, more Solaris, more open source...

At 4:58 PM the software dudes contritely ceded the platform to me, leaving me just 2 minutes to explain our storage strategy. A few years ago I would have been in miserable trouble. But we are building storage from general purpose computers and general purpose operating systems - Sun's x64 and UltraSPARC systems and Solaris 10! And of course, that's what we had just spent the last few hours discussing with this customer.

Granted it took me a few more than 2 minutes to explain the changes in the storage industry that are driving us in this direction. Data is exploding - multiple exabytes of new data being created every year! And the tradition in the storage industry has been to build special purpose hardware and software to manage that data. This custom and proprietary strategy worked because the performance and reliability requirements of storage just couldn't be met with off the shelf computing. Not anymore! When we think about what we need in a storage OS - it's reliability, availability, security, performance, data integrity, ... When I listened to my software colleague talk yesterday about Solaris, it was all about reliability, availability, security, performance, data integrity, ... And a few more cool things thrown in for good measure, like observability (DTrace), virtualization (containers), "better than RAID6" software RAID (in ZFS) and so on. All with a developer community consisting of Sun Storage, Sun Software, and Sun Server engineers, along with our partners and the world participating in our open source efforts. Why would I want anything but Solaris to manage my data?

And what about storage controllers and their processors? We've got them coming out the you-know-what... Fast, cool, space-saving, really sweet storage controllers. We just happen to also call them servers.

People in every division of Sun are working on very interesting storage solutions, which makes it was easy to explain our storage strategy. Good thing, because I hate being the person between a hungry group of people and their dinner.

I'm sure Riley was trying to achieve a state of Zen-ness with his hard work at this garden... look at the dinosaurs intertwined with the soldiers... the piles of rocks... even a pen cap thrown in for good measure... (is that uncapped pen lurking in the cushions of my chair?) And while the contemporary artists among us might relax with the results, the more conventional like me were just exhausted by the time Riley and dad left for the day.

And how is this related to storage you might ask? This is exactly what happens when you take special purpose storage technology from the 90s, running single threaded operating systems on specially built boards, and you put it into today's world. How can you run your Web 2.0 application on your storage system? Once you move aside the dinosaurs and pen caps, you're left with nothing but a mess of firmware trying to handle sectors, slices, and spindles.

That's why we're building our storage systems today with general purpose hardware and software. That's why we think Thumper matters. Because our world today is different. We've got to rake the dinosaurs, pen caps, and soldiers out of storage and give today's developers a better way to manage their data.

Check out the Big is Bad emblem on the tow-hitch - a perfect statement. But this big bad thumper isn't nearly as eco-responsible as our hybrid data storage server at Sun. Our x4500 data storage server holds 24 TB of data in a 4U of rack space, starting at $2/GB. A data storage server that can run any application - right on the storage system itself!

We're building new storage systems like this just because there's too much data in the world. Too much to just shove away on a block storage device and forget about. Too much to simply spool to a tape and ship offsite. Data today matters to more and more people over longer and longer periods of time. Data that has to be accessible when it's needed. You know what that means? It's not so much about how you store it - sure there are tons of options for disk and tape to meet your performance, reliability, scale, etc needs - its really about the applications you use to retrieve it for use again and again...