Last year was a very productive and busy time for me. I spent a lot of time with our CloudForms product, developing best practices and implementation guides for our consultants. I even contributed some of my work upstream in the form of this disconnected sync script and started a video series for katello. This year I plan to complete this series and continue producing video screencasts for other emerging products. In fact, I just completed a screencast on Red Hat Storage which I will post here soon.
However, what I’m most excited about is my recent work with OpenShift Enterprise. What is OpenShift Enterprise? It is a customer installable version of our OpenShift Online PaaS service. I can’t really describe how excited this product makes me. Just check out some of the cool uses on the OpenShift blog such as the recent BrowserQuest post by Mike McGrath.
I will be blogging more frequently so stay tuned for some tips, tricks, and hopefully more screencasts. If you really want a taste of a fully distributed OpenShift architecture, be sure to read Scott Collier’s amazing Reference Architecture which I have the pleasure of reviewing in detail. For a great write-up on using CloudForms to provide the infrastructure for deploying OpenShift nodes see James Labocki’s blog: http://allthingsopen.com/.
Finally, I hope to spend some quality time with Red Hat OpenStack which is available in preview now.
I’m literally soaking my feet right now. I definitely wore the wrong shoes to be working a booth the whole time.
I had a hardware failure on one of my systems for the demo. Why couldn’t it have been one of the easily re-installable hypervisors? No, it had to be the fully synchronized System Engine VM. I considered asking if I could request everyone stop using their WiFi so I could sync the 17GB I needed. Demo fail always tries to get me. But I wasn’t going down without a fight.
I switched to plan B: playing a recorded video while I rebuilt it. It took some elbow grease, but the demo went well. Summit was a blur, but I had some great conversations with customers, coworkers and friends.
I’m really excited about CloudForms. In my recorded demo I showed a RHEL 2-node active/passive cluster with GFS off an iSCSI target. Then I moved all the underlying CloudForms Cloud Engine components to shared storage. I was able to launch instances, fail over Cloud Engine, and see the correct status. After managing the instances, fail back, and all was good. All of this works because the RHEL HA cluster stops the databases and other services first, moves the floating ip over, then starts the services on the active node. This was a very basic deployment, much more could be explored with clustered PostgreSQL and sharded Mongo.
In the demo booth at Summit I showed customers what CloudForms is, how to walk through a typical workflow, and demonstrated launching a 4-node JBoss cluster based on Scott Collier’s great Reference Architecture: http://www.redhat.com/resourcelibrary/reference-architectures/deploying-red-hat-openshift-enterprise-paas-for-itops which I had the pleasure of technical reviewing. I had a local RHEV 3.0 datacenter. I wanted to add a VMware vSphere 5.0 environment too, but it didn’t support the laptop NIC, so I settled for 2 RHEL systems attached to RHEV. I know I’m biased, but the snapshot method RHEV uses is exponentially faster as more instances are launched simultaneously. Boot time of 4 instances (2GB ram each) was under 3 minutes. On my enterprise class lab hardware, just 6 nodes takes about 8 minutes to fully copy and deploy on vSphere 5.0 due to the way each VM is copied from the template. I also had my EC2 account connected, and launched instances in both environments.
So, what exactly is CloudForms? The core focus is on lifecycle management of virtual machines. However, rather than attempting to displace existing virtualization management infrastructure, CloudForms acts as an abstraction layer above virtualization and uses deltacloud to interact with VMware vCenter, RHEV Manager, or Amazon EC2. This means customers aren’t asked to waste any investments in infrastructure. Virtual machines deployed by CloudForms are referred to as instances. One or more instances can be deployed from CloudForms and managed as a single entity and are referred to as applications. The user that launches the applications can choose which infrastructure to launch on. CloudForms has three major components:
- CloudForms System Engine – manages content, image definitions, and software updates – based on the upstream Katello http://katello.org/
- CloudForms Cloud Engine – manages the lifecycle of images, instances, and applications – based on the upstream Aeolus http://aeolusproject.org/
- CloudForms Config Server – provides post-boot configuration for instances – based on the upstream Audrey http://aeolusproject.org/audrey.html
Be sure to check out the great keynotes from Red Hat Summit: http://www.redhat.com/summit/
If you want to get started with CloudForms check out this great overview from Jacob Liberman: http://www.redhat.com/resourcelibrary/reference-architectures/red-hat-cloudforms-v1-architectural-overview
For a deployment guide with use cases, this one by Steve Reichard I also helped review: http://www.redhat.com/resourcelibrary/reference-architectures/red-hat-cloudforms-v1-infrastructure-and-application-deployment-fundamentals
Of course, we’ll help you get it up and going fast: http://www.redhat.com/consulting/
I’ll post a link to my video once I figure out where to put it.
Thanks for everyone who came to Summit. Fenway Park was a blast, the weather in Boston is wonderful compared to Austin.
Update: I’ve uploaded my video until I get get some better intro/outtro on it, at which point I’ll make a separate post: http://people.redhat.com/vvaldez/videos/summit-2012/
 CloudForms 1.0 supports VMware vSphere 4+/5+, Red Hat Enterprise Virtualization 3.0+, and Amazon EC2. Support for more virtualization environments are in the works.
I failed to talk a bit more about myself in my first post. That’s not really a bad thing.
One thing I didn’t mention is that I am very new to blogging, so please bear with me as I settle into my comfort zone.
I have been quite busy lately. My current focus is on our Red Hat Consulting Pathways. This has been a large project, with details to come. The idea is to help customers with larger impact projects. Not just transactional product implementations. The end goal is to help customers go from traditional disparate IT environments usually running some UNIX workloads and helping them map a strategic pathway to a more dynamically configurable enterprise using open source and open standards, evaluating the viability of using cloud resources where necessary.
Now I’m working with our Red Hat CloudForms product, currently in beta. I’m very excited with the versatility of this product, and once it is released I have a bunch of implementation details I can’t wait to share. If you are curious, the beta documentation is now live.
One other item that has kept me busy is a project a colleague and I have started. We want to help budding system administrators learn Linux. We know Linux may be a bit obtuse for those not familiar with a command-line interface, so we are creating screencasts of common Linux commands with voice narration at the CLI Academy. We are largely using the process I outlined in this blog post using ffmpeg for screen capture, Blender for the Video Sequence Editor, and Audacity for audio capture and editing. However, we are making more use of OpenShot for animated titles and any simple editing. We plan to add more complex topics and courses once we build a library of basic commands.
Lastly, I wanted to mention it has been a bit of a challenge to keep track of all of my various projects. I recently read David Allen’s “Getting Things Done” personal task management book. I must say it is a fantastic methodology. It takes some work to incorporate into a daily routine, but I think it is worth it. One quick and simple program I found that uses this process is Getting Things Gnome for Fedora. However, if you fancy a Mac system, OmniFocus is the best I’ve seen. I don’t use an iPhone, but it syncs nicely with an iPad for management on the go. There is a service offered by Spootnik that offers a web interface into OmniFocus that also allows syncing with BaseCampHQ for a kludgy team based solution. For other team task management, we have used trac, but have recently moved to PHProjekt and are really enjoying it.
So … what factors are important to you in migrating from UNIX to Linux and/or preparing for utilizing cloud resources? What task methodologies do you find best for personal and team management?
I’ve always loved puzzles.
I remember taking apart a door knob when I was very young, and my mom wasn’t happy about it since it was the front door. My punishment was to figure out how to fix it before my dad came home.
I have continued solving puzzles throughout my career. Mostly tactical problems that required some creative thinking, like this open source project I started to solve an interesting challenge with managing Dell Linux Repository software through RHN Satellite.
It is interesting what happens when you take someone like that and put them in a more strategic role, where problems have less defined symptoms and linear solutions that can be solved with some code. The last several years at Red Hat I have shifted into this role. Only, my focus is on helping enterprises and partners consolidate, virtualize and begin the pathway to the cloud.
That isn’t exactly an easy road. There are many bumps along the way. I am tasked with identifying those bumps and making a sensible roadmap to getting there.
For example, running multi-tier applications and wondering which components in the stack could be migrated or scaled into a public cloud? Why would you want to do this? How can you get your organization ready and benefit from using different cloud options? Is a cloud right for you? Isn’t a cloud just virtualization with some extra buzz words?
I hope to use this blog to not just talk about myself, but rather, collaborate with my readers and understand what concerns you have. I will still be getting my hands dirty as I work more with our forthcoming CloudForms product, like this video where I show an example workload deployment that is defined once and deployed to different providers using an earlier version of the upstream aeolusproject.
I hope to share my thoughts and lessons I learn as I tackle this much larger and complex puzzle.
Photo credit: Furk@N