Related articles by Zemanta

• How You Can Help the Earthquake Victims (abcnews.go.com)
• American Red Cross Releases $1,000,000 in Aid to Help Earthquake Victims in Haiti (kitsapsun.com)
• Haiti earthquake: How to help (msnbc.msn.com)

Today at ZDNET Mary Jo Foley wrote an interesting piece entitled:
 

Microsoft, HP to unveil 'solutions built on new infrastructure-to-application model'

Seriously,

Microsoft is once again late to the game.

So, they are attempting to make yet another 'empty' promise for future services that don't exist today.

Microsoft would like nothing better than to have CFOs freeze their Cloud development initiatives and wait for them to catch up.


Amazon Web Services is continuing its expansion and has a total of over three years experience in Cloud Infrastructure, Operating System as a service with 'on demand' pay for only the resources you need being its main attractor.

Don't limit yourself to a Microsoft-only proprietary Cloud.

AWS is agnostic and gives you flexibility to choose which O/S to use, including Microsoft Windows Server 2008, without any license, if you so choose.

Amazon Web Services is here today.

Mary Jo was kind enough to reply to my Talkback with:

Hi. I agree with you that Microsoft is chasing Amazon in the cloud.

As an Amazon partner, I'm curious whether you agree or have anything to say about a recent blog post claiming Amazon is hitting a scalability wall. Do you think it's true?

Here it is: http://alan.blog-city.com/has_amazon_ec2_become_over_subscribed.htm

MJ

To which I replied:

There are two sides to a coin, and an edge.

The author of the above-referenced article, fails to mention whether they contacted Amazon, and if so, were AWS technical support able to corroborate any of their issues and/or what remediation may have taken place in the AWS data center to rectify their latency issues.

Image via CrunchBase

Also, it should be noted latency, by itself, does not imply a scalability issue.

For example, if a web app user spawned a thread for a 'poorly-optimized' back-end database SQL query, that query might be contributory to latency seen in an app, if the query doesn't return its results in a timely fashion.

Moreover, SQL correlated queries are preferred over nested dynamic queries and if developers of the web app's database were diligent in testing their SQLs (e.g., using Explain) they'd have readily seen where their DB bottlenecks are.

Something as simple as identifying a slow-running query and observing that it is running a 'brute-force' sequential table scan vs indexed read can be remedied with the creation of a new primary key index on the column or columns of the database which are part of a SQL select/insert/delete query statement.

It is stuff like the above that has hit concerns like Twitter. The tools they were using were first suspect (Ruby on Rails), but it was really a matter of their coming to terms with their own programming and scalability issues that ultimately allowed them to become the success they are today. Their scalability issues were not due to the inability of RoR to scale, it was only that they had to go through their own 'learning curve' on how best to configure their servers for peak demand usage that was at issue.

No fault of the tools/resources--just a 'learning experience'.

So, while I won't rule out the possibility that AWS may have intermittent 'growing pains', I am always concerned when only one side of a story is given in a blog.

To be fair, the author could have given Amazon the opportunity to provide a response which could have been included in the story.

What do you think?
 

Related articles by Zemanta

• EngineYard - Cloud Based Hosting Platform for Ruby on Rails Web Apps (gajeebo.com)
• Winners And Losers of 2009 - III (cloudave.com)
• MegaRoundup: Windows Unveils Azure Pricing (datacenterknowledge.com)

Mary Jo Foley of ZDNET's 'All About Microsoft' blog writes today an article entitled:

Is VMware the real motivator behind latest Microsoft-HP deal?

I am inclined to believe that Red Hat will factor heavily this year and following more than any other player into the growing market for server consolidation using virtual machine technology.

Keep in mind that every copy of the Linux kernel now includes Kernel Virtual Machine for free.

KVM is a Type 1 hypervisor and has performance characteristics as good as or better than VMware ESX.

Couple that to Red Hat's open sourcing SPICE and you can readily see they have positioned themselves to meet the demand for server consolidation, particularly in a recession when the pressure is on IT to find any way possible to curb or even better 'reduce' software and capital equipment expenditure.

One only need watch the video benchmarks here that compare Citrix ICA, RDP, and SPICE to get a feel for the level of competition coming from open source for Virtual Machine/Thin Client software.

That said, 2010 will be another GREAT year for Red Hat.

What do you think?

Related articles by Zemanta

• Red Hat takes on VMware with server virtualization solution (arstechnica.com)
• Red Hat debuts virtualization management (news.cnet.com)
• Red Hat promotes cloud projects, pans Microsoft and VMware (infoworld.com)

If you administer website support and want to stay on top of site service interruptions,  implementing separate servers running ping checks can be a costly endeavour.

A great, cost-effective alternative is now available: Pingdom

If you have critical web applications that simply can't afford to be down and don't want the hassle of setting up your own Nagios, then pay a visit to Pingdom and set up your service checks in just five minutes.

My site is using Pingdom and I must say I know what an effort it is to set up Nagios.

Having Pingdom is a 'no brainer' for those who would rather not get into the 'nitty gritty' of maintaining yet another dedicated server for ping checks.

Add Pingdom to the growing list of 'genuinely useful' Cloud SaaS products.



Check it out Folks!

Dietrich T. Schmitz
Dietrich T. Schmitz & Associates.

Related articles by Zemanta

• Announcement: Pingdom Subscription Winners (sixrevisions.com)
• Pingdom Desktop Notifier (ghacks.net)
• 10 Free Server & Network Monitoring Tools that Kick Ass (sixrevisions.com)

With storm clouds steadily approaching, here are a few things an organization should take into consideration on an individual basis before making a move to the cloud. 
These are some measures needed to be taken in order to appropriately determine your organizations risk whether real or perceived.

First and foremost, I would propose any organization to re-evaluate their business function model (BFM). 

There are two components to business continuity planning (BCP).
1. Business impact analysis - BIA is concerned with evaluating the process and what impact a loss would have on the organization.

Here are some things to consider for a business impact analysis:

• Identify Critical Functions -- What functions are necessary to continue operation in the event of a disruption?
• Prioritize Critical Functions -- What happens if your data communication service is disrupted?
• Calculate A Time Frame For Critical System Loss -- How long of a disruption can our organization withstand?
• Estimate The Tangible And Intangible Impact On The Organization -- What is the public relations repercussions?
  

2. Risk assessment sometimes referred to as risk  analysis - Risk analysis primarily deals with the threats, vulnerabilities, impacts related to the loss of information, processing capability and data availability.

Here are some things to consider for a risk assessment:

• Risk Exposed To The Organization -- Operating systems and applications are known to be at risk in certain environments.
• Risk That Need Addressing --  Resources should be allocated in such a way to prevent theft and espionage.

BCP is a management tool that ensures the BFM can be performed when normal business operations are disrupted.  The risk assessment, in conjunction with the BIA, provides a organization with an accurate picture of the decisions facing it.  It also allows the organization to make intelligent decisions about how to respond to various scenarios.

Before anyone makes a move to the cloud, it would be advantageous to incorporate their business model with a true annual loss expectancy (ALE).
ARO - (annualized rate of occurrence) - Is historical data of occurrences happen within a year.
SLE  - (single loss expectancy) - A monetary value assigned to data.
ARO x SLE = ALE
example:
Lets say an organization's in house web server generates $25,000 in revenue per hour.  The probability of the web server failing is 25 percent, and a disruption causes 3 hours of down time and $5,000 in repair cost.  The SLE is $80,000($25,000 x 3 hours + $5,000), and the ARO is .25.  The ALE is $20,000($80,000 x .25).

From a technical perspective, securing of information will be done by the existing confidentiality, integrity, and availability (CIA) technology used in information technology today, with more enhancements with the focus being on speed of compression/decompression and  encryption/decryption. 

Also expect the US government to implement a key escrow system for the purpose of law enforcement access.

The weak link in on-line security as it exists today is the single factor authentication model which is predominately used today.  However, there is nothing intrinsically wrong with single factor authentication. They would have to at lest go to some form of a multi-factor authentication model to achieve a greater level of acceptance by the industry as a whole and the masses.

In conclusion:  If done correctly, the cloud should only augment one's current infrastructure not replace.

---

Peace and Prosperity to all my Friends and Family in 2010!

Watch me hip hop!

OK,

Very recently, I upgraded to Ubuntu 9.10 Karmic Koala.  I've got no issues with Ubuntu.  Things are fine, only I noticed running htop that only one processor CPU0 was showing utilization.

What's up with that?  Digging in further with:

$ dmesg | grep CPU

yields a key piece of information: Brought up 1 CPUs
OK, lets see what this turns up...

$cat /proc/cpuinfo:

processor    : 0
vendor_id    : AuthenticAMD
cpu family    : 15
model        : 72
model name    : AMD Turion(tm) 64 X2
stepping    : 2
cpu MHz        : 800.000
cache size    : 512 KB
physical id    : 0
siblings    : 1
core id        : 0
cpu cores    : 1
apicid        : 0
initial apicid    : 0
fpu        : yes
fpu_exception    : yes
cpuid level    : 1
wp        : yes
flags        : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt rdtscp lm 3dnowext 3dnow up rep_good extd_apicid pni cx16 lahf_lm cmp_legacy svm extapic cr8_legacy
bogomips    : 1607.22
TLB size    : 1024 4K pages
clflush size    : 64
cache_alignment    : 64
address sizes    : 40 bits physical, 48 bits virtual
power management: ts fid vid ttp tm stc


and let's verify the kernel, shall we?

uname -a
Linux dietrich-laptop 2.6.31-14-generic #48-Ubuntu SMP Fri Oct 16 14:05:01 UTC 20

OK, while it seems odd, let's not jump the gun.  Could this be a regression?  Maybe.
After some googling I found a bug#14525 opened up with the same issues I have experienced.

Just out of curiosity, I asked a friend of mine who lives in Europe to check his dmesg output.
Bingo.  He is running Ubuntu 9.10 x86_64 SMP as well on an AMD Turion64x2 and has only one wing engine running!

OK, time to open a bug report #14593.  Done

If you read the other chap's bug report, he pretty much has demonstrated that there appears to be a regression going back as far as Linux Kernel 2.6.24.

If you have this issue--please amplify and extend my bug#!
Hopefully, I'll have an answer on this issue sooner than later.

--Dietrich

Related articles by Zemanta

• AMD shakes up x86 CPU world with two new designs (arstechnica.com)
• AMD Has A Vision, But Will It Be Right For You? (lockergnome.com)
• AMD's new laptop platform improves on graphics (infoworld.com)

Dear IT Decision Maker,

These are important times.   

Given the economic circumstances, due-diligence requires that alternatives be considered.

Such alternatives include stepping off of the MS Licensing treadmill to FOSS alternatives.

When IBM and Canonical inked an agreement for Ubuntu Desktop for Business (Press Release), you know that there is serious consideration being given to Linux as a replacement for Windows.



A 'wait and see' approach that includes making a full evaluation of all options until at least Service Pack 1 of Windows 7 is prudent while making a decision on where future dollars can be best applied toward capital investment.

Unfortunately, if you are trying to get a 'fair and balanced' view at ZDNET you are most unlikely to get that.

So, Dear IT decision makers keep your eyes wide open and ask the serious question, one not included in these Q&A's at ZDNET:

Do I really need Windows 7 or would our business be better served by Ubuntu Desktop for Business?

I repsect Ed Bott's intellectual efforts put into supporting MS Windows but if he is truly honest, he will provide his readers a chance to consider 'the flip side' and reach their own conclusions.

Ed I challenge you to make a side-by-side feature matrix of Windows 7 vs Ubuntu 9.10.
Are you up to the challenge Ed?

(I put this very question to him directly in the ZDNET Talkbacks.)

--Dietrich

Related articles by Zemanta

• IBM's Answer To Windows 7 Is Ubuntu Linux (linux.slashdot.org)
• IBM launches Ubuntu-based distro in Africa (slumpedoverkeyboarddead.com)
• Mark Shuttleworth Calls Windows 7 A Nice Service Pack (slumpedoverkeyboarddead.com)