Friday Jul 27, 2012

What To Give Your Favorite Sysadmin on Sysadmin Day


Happy Sysadmin Day.

As the site says ...

This is the day that all fellow System Administrators across the globe will be showered with expensive sports cars and large piles of cash in appreciation of their diligent work ..."

We understand that times are tight, so we don't expect you to buy your favorite sysadmin a Ferrari. That can wait till next year. But it wouldn't hurt to rent them one. Just for the weekend. To remind them what a weekend feels like.

- Rick

Website Newsletter Facebook Twitter

Tuesday Jul 17, 2012

How to Protect Your Oracle Linux System from the Higgs Boson

Now that the Higgs Boson particle has been gently coaxed out of hiding, you know what's gonna happen, don't you? Your boss is gonna walk into your office and demand a plan for protecting your Oracle Linux system against it.

You could act like a smart aleck sysadmin and inform him or her that it took a team of scientists 10 years and 500 trillion collisions to get conclusive evidence of its existence, and let's not even talk about how difficult it was for God to create the elusive thing, but that would violate the first law of corporate survival:

Never, ever make your boss look stupid

Instead, jump out of your chair and say "OMG! I hadn't though of that!" Then read our latest article and use what you learn to write up a plan that will make your boss look real good to his or her boss. (Just make sure your name appears nowhere.)

Tips for Hardening an Oracle Linux Server

Lenz Grimmer and James Morris provide guidelines for:

  • Minimizing the software footprint
  • Minimizing active services
  • Locking down network services
  • Disabling or tightening use of SSH
  • Configuring mounts, file permissions, and ownerships
  • Managing Users and Authentication
  • Other Security Features and Tools
  • Cryptography
I hope you enjoy reading the article as much as I did. And good luck with your career.

- Rick

Website Newsletter Facebook Twitter

Thursday Jun 28, 2012

Similar But Not The Same

A few weeks ago we published an article that explained how to use Oracle Solaris Cluster 3.3 5/11 to provide a virtual, multitiered architecture for Oracle Real Application Cluster (Oracle RAC) We called it ...

How to Deploy Oracle RAC on Zone Clusters

Welllllll ... we just published another article just like it. Except that it's different. The earlier article was for Oracle RAC This one is for Oracle RAC This one describes how to do the same thing as the earlier one --create an Oracle Solaris Zone cluster, install and configure Oracle Grid Infrastructure and Oracle RAC in the zone cluster, and create an Oracle Solaris Cluster resource for Oracle RAC-- but for version of Oracle RAC. Even though the objective is the same, and the version is only a dot-dot-dot release away, the process is quite different. So we decided to call it:

How to Deploy Oracle RAC on Zone Clusters

Hope you can keep the different versions clear in your head. If not, let me know, and I'll try to make them easier to distinguish.

- Rick

Website Newsletter Facebook Twitter

Tuesday May 29, 2012

Is Linear Tape File System (LTFS) Best For Transportable Storage?

Those of us in tape storage engineering take a lot of pride in what we do, but understand that tape is the right answer to a storage problem only some of the time. And, unfortunately for a storage medium with such a long history, it has built up a few preconceived notions that are no longer valid.

When I hear customers debate whether to implement tape vs. disk, one of the common strikes against tape is its perceived lack of usability. If you could go back a few generations of corporate acquisitions, you would discover that StorageTek engineers recognized this problem and started developing a solution where a tape drive could look just like a memory stick to a user. The goal was to not have to care about where files were on the cartridge, but to simply see the list of files that were on the tape, and click on them to open them up. Eventually, our friends in tape over at IBM built upon our work at StorageTek and Sun Microsystems and released the Linear Tape File System (LTFS) feature for the current LTO5 generation of tape drives as an open specification.

LTFS is really a wonderful feature and we’re proud to have taken part in its beginnings and, as you’ll soon read, its future. Today we offer LTFS-Open Edition, which is free for you to use in your in Oracle Enterprise Linux 5.5 environment - not only on your LTO5 drives, but also on your Oracle StorageTek T10000C drives. You can download it free from Oracle and try it out.

LTFS does exactly what its forefathers imagined. Now you can see immediately which files are on a cartridge. LTFS does this by splitting a cartridge into two partitions. The first holds all of the necessary metadata to create a directory structure for you to easily view the contents of the cartridge. The second partition holds all of the files themselves. When tape media is loaded onto a drive, a complete file system image is presented to the user. Adding files to a cartridge can be as simple as a drag-and-drop just as you do today on your laptop when transferring files from your hard drive to a thumb drive or with standard POSIX file operations.

You may be thinking all of this sounds nice, but asking, “when will I actually use it?” As I mentioned at the beginning, tape is not the right solution all of the time. However, if you ever need to physically move data between locations, tape storage with LTFS should be your most cost-effective and reliable answer. I will give you a few use cases examples of when LTFS can be utilized.

Media and Entertainment (M&E), Oil and Gas (O&G), and other industries have a strong need for their storage to be transportable. For example, an O&G company hunting for new oil deposits in remote locations takes very large underground seismic images which need to be shipped back to a central data center. M&E operations conduct similar activities when shooting video for productions. M&E companies also often transfers files to third-parties for editing and other activities.

These companies have three highly flawed options for transporting data: electronic transfer, disk storage transport, or tape storage transport. The first option, electronic transfer, is impractical because of the expense of the bandwidth required to transfer multi-terabyte files reliably and efficiently. If there’s one place that has bandwidth, it’s your local post office so many companies revert to physically shipping storage media. Typically, M&E companies rely on transporting disk storage between sites even though it, too, is expensive.

Tape storage should be the preferred format because as IDC points out, “Tape is more suitable for physical transportation of large amounts of data as it is less vulnerable to mechanical damage during transportation compared with disk" (See note 1, below). However, tape storage has not been used in the past because of the restrictions created by proprietary formats. A tape may only be readable if both the sender and receiver have the same proprietary application used to write the file. In addition, the workflows may be slowed by the need to read the entire tape cartridge during recall.

LTFS solves both of these problems, clearing the way for tape to become the standard platform for transferring large files. LTFS is open and, as long as you’ve downloaded the free reader from our website or that of anyone in the LTO consortium, you can read the data. So if a movie studio ships a scene to a third-party partner to add, for example, sounds effects or a music score, it doesn’t have to care what technology the third-party has. If it’s written back to an LTFS-formatted tape cartridge, it can be read.

Some tape vendors like to claim LTFS is a “standard,” but beauty is in the eye of the beholder. It’s a specification at this point, not a standard. That said, we’re already seeing application vendors create functionality to write in an LTFS format based on the specification. And it’s my belief that both customers and the tape storage industry will see the most benefit if we all follow the same path. As such, we have volunteered to lead the way in making LTFS a standard first with the Storage Network Industry Association (SNIA), and eventually through to standard bodies such as American National Standards Institute (ANSI). Expect to hear good news soon about our efforts.

So, if storage transportability is one of your requirements, I recommend giving LTFS a look. It makes tape much more user-friendly and it’s free, which allows tape to maintain all of its cost advantages over disk!

Note 1 - IDC Report. April, 2011. “IDC’s Archival Storage Solutions Taxonomy, 2011”

- Brian Zents

Website Newsletter Facebook Twitter

Thursday May 24, 2012

Overcoming Your Fear of Repositories

One of the best features of Oracle Solaris 11 is its software update model. As you have probably heard many times by now, the Image Packaging System (IPS) handles package dependencies automatically, so you no longer have to check them manually or create scripts that assemble the correct set of packages .

If you don't have a support contract, you have to wait until the next release of Oracle Solaris 11 to get the latest updates. But if you do have a support contract, you can keep your system updated with the latest security updates and bug fixes by downloading updates from the Oracle Support Repository. We recently published two articles that describe how, plus one more that shows you how to create multiple internal repositories.

How to Update Oracle Solaris 11 Systems from the Oracle Support Repository, by Glynn Foster

You may already know that you don't have to worry about manually tracking and validating patch dependencies when you update a version of Oracle Solaris 11. But did you also realize you can apply the updates to an alternate boot environment, and then schedule the switchover to happen automatically at a later time? Glynn Foster explains how, and how to make sure everything went well.

More Tips for Updating Oracle Solaris 11 Systems from the Oracle Support Repository, by Peter Dennis

The Oracle Support Repository contains bug fixes and critical security fixes that can be applied to existing Oracle Solaris 11 installations between major releases. The repository is updated monthly. Peter Dennis describes how to access those updates and apply them to your systems.

How to Create Multiple Internal Repositories for Oracle Solaris 11, by Albert White

Even though you may get all your software updates to Oracle Solaris 11 from an external repository, you may still want to create different internal repositories to serve different versions of Oracle Solaris 11 to different types of systems. Albert White shows you how to create and manage internal repositories for release, development, and support versions of Solaris 11.

There's plenty more where these came from. Be sure to bookmark our Installation Spotlight page, maintained by the kind and prolific folks who bring you Oracle Solaris 11.

About the picture ...
Laird Hamilton is a god.
Teahupoo is a killer wave.
Laird owned it.
Be like Laird.

- Rick

Website Newsletter Facebook Twitter

Tuesday May 22, 2012

Cluster Fest

It's been a couple of months since we focused on Oracle Solaris Cluster. If you're a fan, we have some new content that will interest you. See below. (If you're new to Solaris Cluster, in particular how to use it in a virtual environment, see "Recent Technical Articles About Oracle Solaris Cluster," further down.)

New Technical Articles About Oracle Solaris Cluster

How to Upgrade to Oracle Solaris Cluster 4.0
If you are running Oracle Solaris Cluster 3.3 5/11 on Oracle Solaris 10 and want to upgrade to Oracle Solaris Cluster 4.0 running on Oracle Solaris 11, consider using the Oracle Solaris Cluster Geographic Edition software. It makes the job easier and keeps downtime to a minimum. Tim Read wrote this 8-part article to show you how. Contents are:

How to Deploy Oracle RAC on Zone Clusters
This one is very cool. Oracle Solaris Cluster lets you create clusters of Solaris zones. That gives you high availability. You also get high availability from Oracle Real Application Clusters (RAC). So why would you install RAC on zone clusters? Because you can implement a multi-tiered database environment that isolates database tiers and administrative domains from each other, while taking advantage of centralized (and simpler) cluster administration. This article explains how to do it.

Recent Technical Resources About Oracle Solaris Cluster

Blog: How to Survive the End of the World - Part I
Provides a simple example of a two-node cluster, and provides resources to help you create one.

How to Survive the End of the World - Part II
Changes the 2-node example above into a failover cluster, and provides resources to help you create one.

As always you can find the latest technical resources to help you evaluate, test, and deploy Oracle Solaris Cluster on OTN's Cluster Resources for Sysadmins and Developers

- Rick





Thursday May 10, 2012

Adventures in Flopping

Click an image to expand it.

Flopping in the NBA:

Flopping in soccer:

Flopping in the the NFL:

Flopping in the Data Center:

Next time your boss brushes aside the Chitos wrappers strewn among the coaxial cables with a dismissive swipe of his Wingtips so he can make his way into your office and demand that you install the latest version of Oracle Solaris 11 on 3,000 clients by Monday, you won't have to resort to flopping.

Just tell him that it'll take you all weekend, and then read Isaac Rozenfeld's explanation of:

How to Set Up Automated Installation Services in Oracle Oracle Solaris 11

The Automated Installer in Oracle Solaris 11 is kinda sorta the replacement for JumpStart and a very cool tool. You should learn how to use it. Steps are:

  1. Use the command-line to set up a an Oracle Solaris 11 system to act as an Automated Installer server.
  2. Create an installation service that will be automatically installed on clients.
  3. Test the installation service on the client, using default settings.
  4. Run the installation service again, but with custom settings.

Isaac strolls through them in grand style:

For more information about the Automated Installer and other installation tools in Oracle Solaris 11, see the Oracle Solaris 11 Installation Resources Page on OTN.

- Rick





Friday Mar 16, 2012

Oracle Linux Forum

This forum includes live chat so you can tell Wim, Lenz, and the gang what you really think.

Linux Forum - Tuesday March 27

Since Oracle recently made Release 2 of its Unbreakable Enterprise Kernel available (see Lenz's blog), we're following up with an online forum with Oracle's Linux executives and engineers. Topics will be:

9:30 - 9:45 am PT
Oracle's Linux Strategy

Edward Screven, Oracle's Chief Corporate Architect and Wim Coekaerts, Senior VP of Linux and Virtualization Engineering, will explain Oracle's Linux strategy, the benefits of Oracle Linux, Oracle's role in the Linux community, and the Oracle Linux roadmap.

9:45 - 10:00 am PT
Why Progressive Insurance Chose Oracle Linux

John Dome, Lead Systems Engineer at Progressive Insurance, outlines why they selected Oracle Linux with the Unbreakable Enterprise Kernel to reduce cost and increase the performance of database applications.

10:00 - 11:00 am PT
What's New in Oracle Linux

Oracle engineers walk you through new features in Oracle Linux, including zero-downtime updates with Ksplice, Btrfs and OCFS2, DTrace for Linux, Linux Containers, vSwitch and T-Mem.

11:00 am - 12:00 pm PT
Get More Value from your Linux Vendor

Why Oracle Linux delivers more value than Red Hat Enterprise Linux, including better support at lower cost, best practices for deployments, extreme performance for cloud deployments and engineered systems, and more.

Date: Tuesday, March 27, 2012
Time: 9:30 AM PT / 12:30 PM ET
Duration: 2.5 hours
Register here.

- Rick

Thursday Mar 15, 2012

How to Unlock the Performance of Your Multithreaded Applications

Does the memory allocation of your multi-threaded application look like this?

That's probably because you're using a memory allocator designed for single-thread, single-CPU applications. Those memory allocators use malloc() and free() to reserve a portion of memory for your application, and then release it.

That type of memory allocator can continue to work in an application that ventures timidly into the possibilities of a multi-core, multithread-capable system. However, as the thread count begins to increase, different threads will start to request access to memory at the same time, and the traditional memory allocator won't be able to keep up.

Article: How Memory Allocation Affects Performance in Multithreaded Applications

Rick Weisner explains how memory allocation has evolved over the years, then shows you how to recognize a performance deficiency in your multithreaded application. Then he describes how to use three MT-aware memory allocators to take advantage of the performance promises of multi-core systems:

  • mtmemlock shipped with Oracle Solaris
  • libumem shipped with Oracle Solaris
  • the hoard allocator, publicly available

Stop the madness. Read Rick Weisner's article.

- Rick (Ramsey)





Tuesday Mar 13, 2012

Who the Linux Developer Met on His Way to St. Ives

For some reason I still remember this nursery riddle:

"As I was going to Saint Ives
I met a man with seven wives
Each wife had seven sacks
Each cat had seven cats
Each cat had seven kits
How many were going to St Ives?

The answer, of course, is one. More about the riddle here.

Little did I know, when I first learned it, that this rhyme would help me understand the Oracle Exadata Database Machine. Miss Blankenship, please forgive me:

As I was going to St Ives
I met a man with 8 Oracle Exadata Machines
Each machine had 8 sockets
Each socket had 8 cores
Each core had 2 threads
How many CPU's were going to St Ives?

If your i-phone has hobbled you to the point that you can no longer do simple arithmetic in your head, you can get the answer to that riddle by listening to these podcasts (the first one even provides notes):

Podcast: How Oracle Linux Was Optimized for the Oracle Exadata Database Machine

Turns out that when you use off-the-shelf components to build a NUMA system like the Exadata, you lower your hardware costs, but you increase the software work that must be done to optimize the system. Oracle Linux already had a set of optimizations well suited to this task. Chris Mason, director of Linux kernel engineering at Oracle, describes the process engineering used to optimize Exadata's integrated stack, touching everything from storage, to networking, the CPU, I/O speeds, and finally the application. Great Q&A, too.

Podcast: What's So Great About Oracle's Unbreakable Enterprise Kernel?

It's easy to replace your tired rust-bucket of a Linux kernel with the chromed-out Unbreakable Enterprise Kernel from Oracle, but why would you? Sergio Leunissen, Oracle Vice President, and Lenz Grimmer, blogger extraordinaire, explain why it's worth your time to use the Unbreakable Linux Kernel. Sergio and Lenz explain why Oracle went to the trouble to engineer its own kernel, what's included in Release 2, how it is tested, how it is optimized for the Oracle stack, the close relationship with the Linux community, and what benefits it brings developers and sysadmins.

Where to Get It, How to Use It

As you may have already heard, Release 2 of Oracle's Unbreakable Enterprise Kernel for Linux is now available. Here are some resources to help you get started.

- Rick with Todd Trichler





Friday Mar 09, 2012

My Personal Crib Sheet for the ZFS Storage Appliance

Question: What do the F22 Raptor and the ZFS Storage Appliance have in common?

Answer: They bend time. They compress distance. And they both come with their own simulator.

We recently published some articles about really cool ways to use the ZFS Storage Appliance (see below), so I spent a little time looking into the darned thing. It's easy to find out what the ZFS Storage Appliance does, but more difficult to find out what its components are. What can I yank out and replace? What can I connect it to? And what buttons and levers can I push? Or pull.

So I put together this crib sheet. If you didn't grow up in The Bronx, see wikipedia's definition of crib sheet.

What Have We Published Recently?

What the Heck Is It?

It is Oracle's main NAS system for enterprise environments. In case you don't already know, NAS (Network Attached Storage) is simply a storage system designed to be shared by several servers on a network. Instead of each server having its own storage, which would make sharing files wicked slow, you put all your storage on your NAS system, and let all the servers access it fast. Plus, it's much easier to manage. Shoot, you can even store your boot environments on your NAS system so that if one of your servers dumps core, you can reboot it from the NAS system.

It comes in three variations:

  • 7120 - for small and medium size installations - 3.3 TB to 177 TB raw capacity
  • 7320 - mid-range storage for the enterprise - cluster option - up to 288 TB raw capacity - Hybrid Storage Pools with up to 4 TB of optimized cache
  • 7420 - For virtualized environments requiring multiple data services and heterogeneous file sharing - single or cluster - up to 1.7 PB of raw capacity

What Makes It Special?

  • It's wicked fast (see F22 Raptor, above).
  • It's got management software that makes it easy to administer.
  • Its Hybrid Storage Pool Design recognizes I/O patterns and places data in the storage media that will provide best performance for that data, whether DRAM, flash, or disk.
  • Hybrid Columnar Compression reduces storage footprints for NAS-based databases from three to five times.
  • DTrace analytics help you diagnose performance and networking bottlenecks
  • Fault Management Architecure (FMA) identifies faults and automatically re-routes traffic around them.
  • When you need more capacity, you can add:
    • DRAM, cache, or I/O ports for more resources
    • Disk shelf units for greater total capacity
    • Flash drives for faster performance.
  • You can get it in a dual-cluster configuration for high availability.
  • It provides a variety of RAID protections to balance capacity, protection, and performance requirements of your applications.
  • It's waaaaay cheaper than an F22 and doesn't require all that cryptic back and forth with those moody Air Traffic Control people.

Details here.

What's In The Box?

When I asked, I got the usual "Well, I could tell you what's in the box, but then I'd have to shoot you." Turns out they don't want me messing with it. Or you messing with it. The darn thing is built from off-the-shelf components, but the value-add comes from the way they're tuned to work together. So if you, Mister Curiosity, decide to pop open a terminal and run ssh into Solaris, you'll see a message notifying you that if you continue with your wayward ways you'll void your warranty. Ack! Like the good ol' boys from the Georgia Satellites like to put it...

She said, "No huggee, no kissee
Until you make me a wife."
Oh, my honey, my baby
Don't put my love upon no shelf
She said, "Don't hand me no lines
And keep your hands to yourself."

Here's what you really need to know: It's a specialized server with a processor, memory, and disk drives. Loaded with a highly tuned version of Oracle Solaris and other software goodies. But don't think of it that way. Think of it as remote storage. That's all. A box with:

  • Two types of storage:
    1. Filesystem, such as CIFS, NFS, ZFS, etc.
    2. Block, allocated as a Logical Unit (LUN)
  • Connections for a wide variety of network protocols
  • Two sysadmin toolkits:
    1. BUI (boo! boo!)
    2. CLI (yay! yay!)
  • Analytics to help you monitor its performance.

Connections? What Can I Connect It To?

For starters, you can connect it to the other servers on the network, through the stock Infiniband HCA's. That's part of what makes it wicked fast. But you can also connect it to other devices through industry-standard network protocols, including:

  • Infiniband
  • Fibre Channel
  • NFS
  • Common Internet File System (CIFS)
  • Internet Small Computer System Interface (iSCSI)
  • NDMP (Network Data management Protocol so it can participate in remotely-coordinated automatic backups
  • A Virus Scan Service
  • NIS naming, LDAP directory, and Microsoft Active Directory services for centralized management of users, groups, hostnames, etc.

What Administration Tasks Does It Require?

Details vary by model and your needs, but basic administration consists of:

  • Defining the storage allocated to each server
  • Making it available to the servers (sharing)
  • Migrating data
  • Integrating it with other applications
  • Taking snapshots
  • Monitoring performance with DTrace Analytics
  • The usual backups, diagnostics, and housecleaning tasks for any server or storage system

Any Examples of What To Use It For?

Turns out you can do lots of cool things with the ZFS Storage Appliance. A partial listing:

For More Information

The Best American Country Song of All Time?

Is it the best? That's debatable. But it's certainly one of my favorite renditions of a country song, from one of my favorite movies of all time.

- Rick Ramsey (with special thanks to Andrew Ness)





Tuesday Feb 28, 2012

Santa Clara On April 10 - Next OTN Sysadmin Day

Before the part of Oracle that was then Sun Microsystems moved in, the facility used to be known as the Agnews Insane Asylum. Some of us who worked for Sun at the time thought the image was hilarious. Some thought it was insensitive. Some believed it was a statement about the rise of the corporate state and the demise of benign government. That was the Santa Clara campus back then, a diverse, magical workplace full of people who held strong opinions about everything, yet managed to have a great time together.

Another topic that incites strong opinions among good friends is Oracle Solaris vs Oracle Linux. Which one is better? Which one should I use? Which one should I learn how to use? At our OTN Sysadmin Days, we let you decide. Pavel Anni always opens our OTN Sysadmin Days with a talk about Oracle's dual OS strategy. He explains why Oracle offers two operating systems, and summarizes the main features of each one. Then we split off into two different groups to get our hands on each OS.

One group gets their hands on the ZFS filesystem, virtualization capabilities, and security controls of Oracle Solaris.

The other group gets their hands on the package management tools, services, and runs levels of Oracle Linux, plus its volume management tools and the Btrfs filesystem.

The truly adventurous sysadmins jump between groups. Both groups learn by doing, using the hands-on labs similar to those on OTN's Hands-On Labs page. Why attend an event in person when you could simply work the labs on your own? Two reasons:

  1. Since you are away from the obligations of the data center, you get to focus on working the labs without interruption.
  2. You get help from Oracle experts and other sysadmins who are working on the same labs as you.

I've been to all our OTN Sysadmin Days so far. The sysadmins and IT managers who attended told me that it was time very well spent. However, our attendance has been low. Not sure whether we haven't gotten the word out to enough people, or whether it's just difficult for sysadmins to get away. In any case, if we don't improve attendance, we'll have to cancel OTN Sysadmin Days.

So if you're interested, register now. Santa Clara on April 10 may be your last chance. The event is free. Here's the agenda:

Time Session
8:00 am System Shakedown
9:00 am Oracle's Dual OS Strategy

Oracle Solaris Track

Oracle Linux Track

10:00 am HOL: Oracle Solaris ZFS HOL: Package Management and Configuration
11:30 am HOL: Virtualization HOL: Storage Management
1:00 pm Lunch / Surfing OTN
2:00 pm HOL: Oracle Solaris Security HOL: Btrfs filesystem
3:00 pm Presentation: Oracle Enterprise Manager Ops Center 11g
3:30 pm Presentation: Setting Up and In-House Development Environment with Oracle Solaris Studio
4:00 pm Discussion: What are the most pressing issues for sysadmins today?
5:00 pm We all go home

- Rick Ramsey





Tuesday Feb 21, 2012

How I Explained Network Virtualization to Bikers

Back when we first launched Oracle Solaris Express, I stumbled upon a couple hundred bikers who were building a bonfire in the woods with the timber from an old cabin, a dozen cans of gasoline, and a couple of Honda Priuses. To avoid a beating, I convinced them to let me explain how virtual networks work. They set down their gasoline and I rescued some upholstery from the fire.

In the good old days, I explained, a proper biker had only one bike, a hardtail Knucklehead with a kick-starter, 5" over forks, and apes with purty leather tassels fabricated from the remnants of a favorite biker momma's chaps. And one leather jacket. Well worn. Naturally, that proper biker wanted to go to many rallies. But because he only had one bike, he could only go to one rally at a time. And he wore the same jacket to each rally. I suggested they call that favorite leather jacket Solaris, and that hardtail knucklehead a NIC.

"Nick," they asked. "Who's Nick?"

"Well, N-I-C," I explained. "It's short for Network Interface Card."

That made them a little restless, but I quickly added that as a result of the one-jacket, one-bike rule, life was good, pipes were loud, and America ruled the world.

They liked that. I got several pats on the back.

Fast forward 50, maybe 60 years, I explained while drawing the diagram above, and now we call ourselves motorcyclists. We have multiple bikes. And they are all EPA-compliant. And in keeping with the sartorial splendor of the court of Louix the XIV, we have one outfit for each bike. I asked them to pretend that each outfit was a zone, and each motorcycle was a virtual NIC, or VNIC. They got restless at the mention of Nick again, particularly after I brought up France, but I held up a well-manicured hand so they would allow me to elaborate. When modern motorcyclists like me want to go to Sturgis, I explained, we get into our Sturgis zone (a 5-day shadow, leather chaps, and obligatory bandana), and throw a leg over our Sturgis VNIC (a blinged-out CVO Harley Davidson 110" Ultra Classic with the dual-tone paint job). When we want to go to Americade, we slip into our Americade zone (a clean shave, a heated vest, and a reflective yellow Aerostich waterproof suit with 10 large pockets), and hop onto our Americade VNIC (a BMW K1200LT with heated seats, cup holder, and GPS). And so on. One outfit for each motorcycle, one zone for each vnic.

That's as far as I got. They gave me a beating and tossed me, my Vespa, and my modster jacket into the lake.

I decided to get some help.

Nicolas Droux, who was part of the engineering team that developed network virtualization (project Crossbow), agreed to explain all this to me. After assuring me that he was not a biker, we got on the phone. And we turned our phone conversation into a nifty podcast.

Podcast: Why and How to Use Network Virtualization

This podcast is easier to absorb if you listen to it in two parts, each about 15 minutes long.

In the first half, Nicolas explains how the process of managing network traffic for multiple Solaris zones across a single Network Interface Card (NIC) naturally led to the development of virtual NICs. And then to the network-in-a-box concept, which allowed you for the first time to create complete network topologies and run them within a single host to experiment, simulate, or test.

In the second half, Nicolas provides more details about combining zones and VNICS to create a test environment. He explains how you can create a zone to function as a virtual network router, for instance, or a virtual load balancer. By isolating these network functions into zones, you can test how your application performs with different settings, and use DTrace to follow the application calls as they are routed through your virtual network. Once you have the optimum settings for the network and the application, you can deploy it in your data center.

Here are some more resources to help you understand network virtualization:

- Rick Ramsey





Monday Feb 13, 2012

OMG! What Did I Just Install?

Quick Quiz:

Q: What's in this Solaris 10 package: SUNWlibstdcxx4S?
A: You cheated. You googled it and found the link to the Oracle Solaris 10 documentation.

You're in luck, because each release of the Solaris 10 documentation contains a Package List for that release. For example:

Now try this:

Q: What's in this Solaris 11 package: compress/p7zip?
A: buzzer!

The Solaris 11 documentation does not include a package list. You can find mentions of some packages through google, but it's hit and miss. And you still don't get the rest of the info about the package that the Solaris 10 documentation included. So how do you find out what Solaris 11 packages you just installed? Here are two methods.

The pkg list Command

The pkg list command lists all the packages currently installed on your system. If you use it, redirect the output to a file so your screen doesn't wind up looking like a scene out of The Matrix. Since package naming is hierarchical, you are likely to find similar packages grouped together in the list. For example:

$ pkg list

You can just list a subset of the packages you are interested in:

$ pkg list driver/network/ethernet/*

By the way, to list all packages that are available for you to install, add -a to the pkg-list command. This example asks for the name of all the packages you can install in the editor group:

$ pkg list -a editor/*

One you have a list of the packages, you can use one of the commands below to get additional info about each package.

The pkg info and pkginfo Commands

The pkg info command provides detailed information about a particular IPS package. For example:
$ pkg info p7zip
          Name: compress/p7zip
       Summary: The p7zip compression and archiving utility
   Description: P7zip is a unix port of the 7-Zip utility.  It has support for
                numerous compression algorithms, including LZMA and LZMA2, as
                well as for various archive and compression file formats,
                including 7z, xz, bzip2, gzip, tar, zip (read-write) and cab,
                cpio, deb, lzh, rar, and rpm (read-only).
      Category: System/Core
         State: Installed
     Publisher: solaris
       Version: 9.20.1
 Build Release: 5.11
Packaging Date: Wed Oct 19 09:13:22 2011
          Size: 6.73 MB
          FMRI: pkg://solaris/compress/p7zip@9.20.1,5.11- 

Here's another example:

$ pkg info -r solaris-large-server
          Name: group/system/solaris-large-server
       Summary: Oracle Solaris Large Server
   Description: Provides an Oracle Solaris large server environment
      Category: Meta Packages/Group Packages
         State: Not installed
     Publisher: solaris
       Version: 0.5.11
 Build Release: 5.11
Packaging Date: Mon Feb 06 22:33:56 2012
          Size: 5.45 kB
          FMRI: pkg://solaris/group/system/solaris-large-server@0.5.11,5.11-

The pkginfo command does the same for any SVR4 packages you may have installed on the same system.

For More Information

- Rick Ramsey with Alta Elstad





Thursday Feb 09, 2012

Zoning Out

So much virtualization. So little time.

You can virtualize your OS ...

You can virtualize your network.

You can virtualize your storage.

Your server.

Even your highly-personalized desktop.

Me? I would like to virtualize my virtualization technologies. I want ONE server. With ONE OS. And ONE toolkit. That can actually be made up of hundreds or even thousands of virtual OS instances, networks, storage devices, desktops, aircraft carriers, or whatever they virtualize next.

You can't quite do that yet, but in Oracle Solaris 11 you can create zones that are easy to clone on other systems. That's a step in the right direction, I think. The following article describes how. In case you're not too confident in your ability to juggle zones, I added an article that helps you get started with zones in Oracle Solaris 11, and a link to more resources.

How to Configure Zones in Oracle Solaris 11 for Easy Cloning

The easiest way to create a bunch of zones is to clone them from one or more originals. That seems simple enough if you are going to clone them on the same instance of Solaris, but what if you'd like to clone them on other systems? In that case, you need to use virtual networks. You need to set up an entire network topology of servers, routers, switches, and firewalls that you can clone right along with the zones. Jeff McMeekin describes how.

How to Get Started Creating Zones in Oracle Solaris 11

If you used zones (containers) in Oracle Solaris 10, you'll appreciate this article. Because zones are more tightly integrated with the architecture of Oracle Solaris 11, they're easier to set up and manage. In this article, Duncan Hardie demonstrates how to perform basic operations with zones: first, how create a single zone using the command line, then how to add an application to a zone, and finally how to clone a zone.

More Zones Resources

  • Solaris 11 Virtualization Page - Links to demos, podcasts, technical articles, and more resources to help you understand zones and how to use them.
  • Zones Collection - See what zones-related content we've published (or found) since the dawn of time.
  • RSS Feeds Page - Subscribe to zones-related content through your favorite reader.

- Rick


Logan Rosenstein
and members of the OTN community


« November 2015
Blogs We Like