Distance vMotion = Stretched Cluster? - blog.scottlowe.org - The weblog of an IT pro specializing in virtualization, storage, and servers

Long distance vMotion still need Layer 2 environment.
If the long distance site is using different IP subnet, it could use MPLS to extend them as one Layer 2 network. It is too difficult for me to implement such kind of environment.

Distance vMotion = Stretched Cluster? - blog.scottlowe.org - The weblog of an IT pro specializing in virtualization, storage, and servers

UCSM 1.4 : Direct attach appliance/storage ports! | The Unified Computing Blog

Good article describes a method to save 10Gbit Switch while deployment few 10Gbit port for FCoE Storage.

UCSM 1.4 : Direct attach appliance/storage ports! | The Unified Computing Blog

RHEL 6.1, cloud platforms, and a new openness

Red Hat Summit recap: RHEL 6.1, cloud platforms, and a new openness - News - Linux for Devices

FW from Linux for Devices RSS

--

RedHat still insist to open their source in Cloud area, and call others(Google, Amazon) to join this.

PS: Always see BMC appearing in cloud platform anywhere.

Test to write blog from iphone app

Just try app on iPhone to publish blog via iPhone application. It is more easy than web browser.

NetApp - Pseudo Benchmark: FCoE

NetApp - Pseudo Benchmark: FCoE

I am surveying the best storage protocol which can be used in VMWare vSphere. I think the best choice is FCoE when combined with Cisco Nexus Switch.

Virtualization technology - part1

It is long time I haven't update my tech blog. Recently I moved to new challenge job for IT service consultant. And work very hard to study and compare existing virtualization product, like VMware vSphere, Citrix XenServer and Microsoft Hyper-V.
I have a little experience to use Citrix Xenserver before work for this new company. The main reason to use Citrix Xenserver is that it provides free edition version for normal feature, so I use this to see what it can do for server management.

After studying the VMware vSphere key feature like vMotion, SvMotion, HA, FT, DRS, VDR, etc.., I really realize the business continuity is so many important for company.

Now I start to evaluate such feature by my Mac pro, and use VMware fusion as VM environment to install ESX, vCenter, Openfiler to simulate the real architecture like enterprise environment. It is so excited to try such feature.

Now I also understand Virtualization is not equal Cloud, but many Public Cloud use virtualization technology to build up the infrastructure. For example: Amozon EC2 use Xen, Chunghwa Telecom HiCloud use VMware. Using Virtualization technology to deployment server farms is very simple and easy, and can provide small to middle enterprise continuity server service with SLA. Why I said small to middle enterprise? Since virtualization is to run several guest OSs in one real Host, the main reason is software limitation, so the usage will not reach the hardware limitation when you only run one OS. You can check the detail link here.
The easy way to optimize the full hardware is to run several OSs in one hardware, so the virtualization technology is the right solution for this situation.

Finally the most important key point is you must estimate your existing application needs how many resources(CPU, Memory, DiskIO, Network traffic) when you want to move it into virtualizaiton environment. So the next task I will plan to build up a standard procedure to monitor existing application server and write some script to test the benchmark value when it reach some peak concurrent user. The value is very important when you want to know actual CTO/ROI.
And what is the next step when one VM is not enough to serve large peak connection. If your server need to scale up, this need more investigating to understand where is the bottleneck. I won't say too much here, may create another log next time. Here my simple suggestion is to use load balancer in the front end if the bottleneck is at web server side.

I also know another open source solution called Hadoop which you can use low cost hardware to  build up Cloud computing, storage. By using this technology you need to rewrite your application  by MapReduce API, it may be a lot cost to migrate your application. It would be a different topic to discuss in my planning.

Remote debug Android by adb tool

The default Android adbd is only listened on 127.0.0.1:5037, so if the USB is not working on developing board, you can't use adb to do something.

The default setting on device, only listen 127.0.0.1:5037


# netstat      
Proto Recv-Q Send-Q Local Address          Foreign Address        State
tcp        0      0 127.0.0.1:5037         0.0.0.0:*              LISTEN
tcp        0      0 192.168.1.128:56559    74.125.47.109:993      ESTABLISHED


Following is the step to enable the remote adbd

# mv /dev/android_adb /dev/android_adb.bk
# setprop persist.service.adb.enable 1
# setprop service.adb.root 1
# setprop ctl.stop adbd
# setprop ctl.start adbd
# netstat

Proto Recv-Q Send-Q Local Address          Foreign Address        State
tcp        0      0 127.0.0.1:5037         0.0.0.0:*              LISTEN
tcp        0      0 0.0.0.0:5555           0.0.0.0:*              LISTEN
tcp        0      0 192.168.1.128:56559    74.125.47.109:993      ESTABLISHED


The Host machine which want to connect to remote Android device, you can use following command:
Ubuntu$export ADBHOST=192.168.1.128
Ubuntu$./adb kill-server
Ubuntu$./adb devices
Ubuntu$./adb shell