Code changes at Google are deployed and picked up eventually by applications, this can take up to a few weeks according to distingused google engineers.

Example: The search result count for different server locations are different. These changes are picked up eventually and can take up to a couple of weeks. ( try it with your own location ip ) 

Los Angeles    356,000,000   Results
http://74.125.47.103/search?hl=en&q=iphone

New York        225,000,000   Results
http://66.249.93.147/search?hl=en&q=iphone

San Francisco  355,000,000 Results
http://64.233.167.99/search?hl=en&q=iphone

What does this mean for enterprise applications running on the GOOG cloud, i.e. Visa financial transaction? You work in Silicon Valley, Shanghai or Madrid and get paid on Monday. Your bank account has been updated in your home town. Take a plane to New York for a sales meeting when you arrive you go to the nearst bakery and want to buy a coffee, sorry insuffcient funds, appengine is still collecting changes on servers for updates please come back in 2 weeks.

I hope this helps cloud consumers get a better idea where clouds stand and what solution fits best without all the smoke and mirrors. In engineering you need to know the break point of every system so that break point never happens in production.

Just my 2 cents,

JM

IBM Autonomics director heralds user facing applications

Google finds no privacy on private roads

Yesterday I visited the Linux Event in San Francisco to catch up on what's new in  Linux and also how cloud computing is changing the Linux crowd. The interesting things for me were the  Ice Cubes(TM) that hardware vendors are now offering. [here]

Also interesting are how iDataPlex(TM) can acutaly generate cooling  rather than cosume and add to an already overgrown carbon footprint. [here]

 I think the main difference and visions of the vendors was obvious. Just walk into the cube solutions; nice and and neatly stacked with air conditioning rushing at you like a tornado just hit. You can  feel, hear, smell, touch and even taste the cooling rush and addtional carbon footprint. The benefit is to create remote datacenters on the fly and to spread the environmental impact accross multiple locations quickly.

 

The green side of System Z is hands down the best solution I've seen. Just run all enterprise SaaS apps on a single box and your good. Since cloud computing is becoming more mainstream most small-medium and enterprise business are now joining and understand the value of Utility Computing to meet critical SLA's for the core business processes.

 There was also a nice session to reimage a used desktop with Ubuntu which was kind of cool. The used boxes where then donated to school for education purposes.

Overall I think the session was good. Very light compared to Java One or Tivoli Pulse in Orlando this year. Is the applicaiton layer becoming more finer grained and optimized since the importance of the underlying operating system is becoming less and less important? Who knows maybe clouds do take affect on which conferences are heavly visited.

• AT&T follows Google and IBM, unveils 'cloud computing
• Cloud Computing and IBM: The pendulum swings back
• SYS-CON launces Cloud Computing Journal
  

 

Enterprise applications may not run on internet centric clouds but rather on utility computing providers for enterprise class quality of service. The zdnet video interviews below describes the challenges and solutions for the next generation of IT datacenters. Is your core business process ready to provide actual service level agreements to customers? Talk utility computing not cloud cloud computing. Hope this helps. 

 

  Cloud computing is fantastic especially looking at how far Google has come in efforts to build an enormous cloud. Who wouldn't want to leverage GOOG's PaaS? Or throw enterprise apps onto it. Unfortunately / fortunately let's get back to reality.

We may not actually run production enterprise applications on an internet centric cloud for critical business applications. The nice Face book App, Spam App, Twitter and WebShop was nice although it is time to get real. 

The good news is that the on-demand model has been available for a while in utility companies and large financial institutions. It's called z/OS on the IBM Mainframe which includes an OS built-in enterprise database DB2 that has been running secure and highly available transactions for decades.

Cloud and Mainframe release management for applications are different. Update a mainframe mission critical application that runs on a single box and your enterprise is up-to-date across all 256 CPU's. This doesn't work with the cloud. According to GOOG distinguished engineer's applications pickup changes eventually over days or weeks. This is unsatisfactory for mission critical applications.

Here are some IBM examples:http://www-03.ibm.com/systems/z/success/testimonials/index.html

Here are some Google examples:http://appgallery.appspot.com/

Enough smoke and mirrors, if you have GOOG stock it may be time to go short.

Just my 2cents.

 

JM

http://camsolutionsinc.com/blog

Monitoring-as-a-Service(TM)

I think SNMP is great for network diagnostics although for providing incidents for event management and autonomic computing it's not helpful. It makes sense to have a protocol that can move easily across clouds ( EC2, Google App Engine, Flexiscale, ... ) , IT departments and mobile  networks.

A lot of IT Systems Management guys today have the challenge of tying all of the events together from different departments for event management and autonomics. Imagine having to tie events not just from three or four world wide distributed data centers but also from applications hosted in a cloud that can be anywhere. Constantly updating SNMP changes and a variatry of different protocols for each vendor who took just enough time to implement monitoring. Playing catch up, for event management and autonomics, may not be possible when leveraging the scalability the cloud has to offer.

Some vendors think it may be a little to early to think about standards or are just assuming IT Systems Management guys with keep the "same old same old" in the cloud with SNMP and other department specific diagnostic protocols. Which may just be the case since it's been that way forever, although I believe this is the perfect opportunity to change and improve how we identify problems wherever
our applications in the cloud are running.

We all know what security issues snmp brings to the table if someone sends 10,000 requests to a network device in a Denial of Service Attack the SNMP reciever today cannot process all the snmp output and the 10,001 request enters unseen into the network. We are looking for qualified events for event management and autonomics from each group ( transaction, web, app, database and network ). Sure JMX, SNMP, SOAP, etc still help for diagnosics of the different groups but when tying all the events together and making changes I think a standard protocol will help.

I am just a monitoring person who is interested to help make positive change to get applications into the cloud and more manageable when more and more business applications get here, since these tasks will be on the IT management guys and my plate anyways.

Hope this helps.

JM

Depending on the application, servers below the enterprise average indicate an oppportunity for deploying the hosted applications to a cloud to reduce cost and incidents of your IT Operations. An example would be your CRM, it's not your core business application and has a few internal users. The cost of hosting the application internal is greater than hosting it at a CRM saas specialist.

The CPU average charts displays all servers in the enterprise. The servers average CPU utilization are scattered across the graph. The orange line is the overall CPU average of all servers. All servers below the orange line indicate the CPU utilization is below the enterprise average. All servers above the enterprise CPU average line indicate that the servers is above the enterprise average.  

There is a company that specializes in architecture, deployment for enterperise  applications to the cloud. Heres the link .

Hope this helps.

Traditionally hosting your monitoring environment in-house with nagios, zenoss or hyperic came with the requirement to have addtional servers. That is not only a burden on your IT budget but also for the environment.

Since monitoring is not your core business process it's pratical to find a monitoring solution that has the most technical deep-dive agents for problem determination and diagnostics as well as beneficial for the environment.

Searching for an optimal solution, software-as-a-service has proven we can help keep the data center footprint practical i.e. SalesForce.com. Here is the Q&A for monitoring-as-a-service

The most challenging and cpu intesive process in monitoring is the correlation of events. Think of a network with 128,000 events per second *. This is not to uncommmen just look at NMS, firewalls and other network components. These events are not all critical or severity one issues, most just informational. The problem is filtering, correlating and escalating the event that brings your service down. Just as important are undetected security breaches by filtering network events.

To correlate the high volume of these events or to detect security issues can be a challenge. Can your 1, 2 or even 10 server in-house hosted monitoring environment process all network, transaction, application, database and systems events? Do you filter 90% or more of the network events in your environment?

Leveraging monitoring-as-a-service infrastructures designed for cloud computing to process and corelate events real-time, may be an alternative to your open source canned solutions. Typically opensource solutions require your businesses resources for patches, problems and fixes. If your core business process is not monitoring why spend so much time and energy maintaining cool GUI software?

more