Cloud-controlled Wi-Fi: Growing a network with SaaS

Takeaway:  Looks at one of the players in the cloud-controlled Wi-Fi space. For SMBs, it’s possible to grow a global network and manage it from the cloud.

Aerohive Networks sell wireless LAN access points and provide centralized control using cloud-enabled control panels. Buy a few little boxes, rent a cloud service, and you have an enterprise network.

But why? I’ve got a little box with antennas in my house and I don’t need centralized control. Okay, a domestic network is nothing like an enterprise network, but what does Aerohive do for Wi-Fi?

Network history, from protocol wars to captive web portals

Networking has radically changed over the decades. The 1980s was the decade of the protocol wars. TCP/IP was around, but it wasn’t a contender. Governments backed the OSI model, all DEC computers talked DECnet and IBM were pushing SNA. Connecting computers with low-level networking technologies like token ring and thick Ethernet was as much fun as herding cats. The 1990s was about office flood-wiring, providing Ethernet ports to cubefarms. The 2000s spread theWi-Fi brand and Wi-Fi hotspots around the world (IEEE 802.11 standards and WLAN Access Points weren’t cool enough).

Now the Internet and wireless protocols are universal we can concentrate on the clever stuff –making networks more flexible, more secure, and more accessible. Office Wi-Fi enables guest access, conference Wi-Fi enables real-time tweets and shopping mall Wi-Fi enables your friends to save you from a disastrous fashion purchase. When you use your BYO device to connect to the local Wi-Fi hostspot, authenticate to a captive web portal, and browse the web, you are reaping the benefits of decades of networking improvements. Getting a device onto a network is no longer the hard part.

The old enterprise wireless LAN

Here’s how a basic enterprise wireless LAN works.

The company attaches WLAN access points on the LAN. The company issues every employee with an approved device. Each device contains MDM (Mobile Device Management) software. The access points allow the approved devices onto the LAN and security software in the network talks to an MDM agent on each device. Simple.

The new enterprise wireless LAN

Here are a few situations where the simple approach doesn’t work.

The enterprise goes BYOD. Employees use their own consumer devices and they don’t contain any MDM software. These devices need secure access to resources on the corporate network, without relying on MDM software. Aerohive options include multiple SSIDs, captive portal, network credentials, and MAC checks.

Visiting consultants can use the network, but only to get to the Internet. No way are they being allowed anywhere near corporate data stores. The guest network can be separated using VLANs, and Aerohive firewalls stop other traffic.

All the salespeople use Apple products. They expect to be able to print and use the projector to Powerpoint people to death. They haven’t got a clue what multicast DNS is, and they don’t care about firewalls unless they get in the way. Which will cause swearing. An Aerohive Bonjour gateway handles the zero-conf networking and connects iPads with Apple services like AirPrint and AirPlay.

Everyone’s device contains a shiny new 802.11n radio, except the boss who uses an 802.11b laptop from the Ark. If the network goes at that pace, everyone (except the boss) suffers. Aerohive adjusts the data rates for each device.

Aerohive hardware and software

The products are basically access points attached to networked computers that run HiveOS (Aerohive’s own operating system). They are less dumb terminals linked to a central machine and more units that carry out their orders.

Aerohive sells a range of boxes for different environments.

• The AP170 is ruggedized for the great outdoors.
• The AP110 is a small box with no visible antenna.
• The AP350 has so many antennas it looks like a robot spider.

The central control comes in the form of a SaaS control panel called the Cloud Services Platform. An enterprise administrator can manage the network using a web browser. (See the Configuration tab below in this screenshot on the Aerohive website.)

The new competitive advantage

Getting a device onto a network is no longer the hard part. The company that comes up with better features, solves problems, or replaces complicated configurations with simple solutions can find global success (at least until Cisco notices and assimilates the business — resistance is futile).

Networking technology keeps changing. The building industry has been around a lot longer than the IT industry, and it does not regularly make all its materials obsolete. Just as buildings have a foundation, skeleton, cladding, and fittings, the Internet has a hardware foundation, OS skeleton, and application fittings (Okay, it’s not a great analogy - if you can come up with a better one, please feel free to add it to the comments below). Anyone walking around with a smartphone in their pocket is carrying hardware, OS, and applications that did not exist a couple years ago.

An enterprise can’t rely on unmanaged wireless devices to meet the needs of its staff. It needs to offer access from far-flung branch offices and central offices, to people with many needs and devices. Aerohive offer flexible devices that can deal with the clever stuff like firewall policy, tunnels, and VPN

Read More

The CloudOS arrives: Map your cloud journey with real tools

Takeaway:  Lays out the benefits of Microsoft’s CloudOS, which consolidates all the tools needed to manage a modern hybrid cloud platform.

With System Center 2012 Service Pack 1, Microsoft delivers the CloudOS. A new label that may take off — CloudOS refers to viewing all the parts of a modern hybrid cloud as one platform. This means thinking of your on-premise Microsoft private clouds, partner-hosted services, and Microsoft-hosted public cloud subscriptions as a single resource pool for running your enterprise software workloads in their most economical and best-performing locations. System Center 2012 Service Pack 1 (SP1) adds significant cloud integration to every component in the System Center suite.

The concept of CloudOS is that you have greater freedom to architect solutions that take advantage of the geography and economies of your business, partners and suppliers, and customers. You can locate application components where they cost less and/or perform better:

• For scaling out and right-placing: If you don’t already have a private cloud resource in the right place or at the lowest price, you can locate a partner in the appropriate region or industry specialization, or subscribe to public cloud services that match the requirements.
• For getting started: Organizations just beginning a cloud journey should take a look at CloudOS (Windows Server 2012, System Center 2012 SP1, and Windows Azure) as a platform that can safely and confidently transition some or all IT services to the cloud. The cloud journey can be on a comfortable, even extended, timetable that is driven by evolving business needs, not by the priorities of legacy IT infrastructure investments.

An economy of scale occurring is that Microsoft is sharing lessons learned in Windows Azure in “trickle down” fashion. That is, the scripts (like PowerShell), and user interfaces (like the Azure Service Management Portal) that manage millions of Windows servers in Azure very efficiently can deliver the same efficiencies, just on smaller scales, in your data center or a service provider’s — or both!. You can essentially create a “mini-Azure“on premise, or share a partner’s “mini-Azure”, or subscribe to pieces of the “real Windows Azure,” and use the same tools across all three clouds.

Connecting the clouds: From the application up

A challenge for enterprises trying to plan long term IT strategies these days is the difficultly in forecasting a solid five-to-ten year roadmap that transitions from a traditional on-premise IT plant to an optimized hybrid cloud environment. Most organizations understand by now that unless they begin to adopt appropriate cloud technologies, they may be at a disadvantage in the global marketplace to competitors that successfully leverage cloud economics.

With CloudOS, Microsoft is creating a tipping-point moment that could accelerate cloud adoption in many companies. There are several components to this success story:

• Microsoft “owns” the whole CloudOS stack, i.e., Windows Server + System Center + Azure.
• System Center 2012 SP1 components create sticky application-layer attach points to the cloud.
• The cloud migration journey need be neither all-or-nothing nor arbitrary: the migration can be modular and application-driven.
• Service providers to populate the hybrid cloud ecosystem have it easier, being able to start up hosted services using lower-cost commodity hardware and at smaller scale.

Each of these components is unique in the industry and they combine to create something tangible: Stepping-stones for an organization from where it is today to an eventual optimized cloud environment, either hybrid cloud or all-public cloud. I heard a CIO remark that CloudOS was real enough to chart out a 10-year roadmap for their company — this was the first set of comprehensive cloud tools and technologies that they could understand, see and touch, and confidently envision with.

Mapping essential cloud characteristics to System Center features

I previously wrote about the U.S. federal NIST’s definition of the essential characteristics of a cloud. It makes sense to evaluate product features against business benefits that are backed up by process validation such as ITIL and the NIST definition. For example:

• The business benefit is “avoid overcapacity,” such as purchasing more infrastructure than you need, than having an idle capacity situation.
• The NIST essential characteristics that enable this benefit are”resource pooling” and “rapid elasticity.”
• CloudOS expresses these characteristics in connectors between System Center Virtual Machine Manager (VMM), Operations Manager, and Orchestrator that can avoid overcapacity situations.

Here are a few specific features in System Center 2012 SP1 that power the CloudOS, mapped to NIST cloud essential characteristics:

1. On-Demand Self Service: App Controller lets users provision Virtual Machines and cloud services. Service Manager lets users open tickets which may invoke Orchestrator workflows that perform provisioning and scaling tasks.
2. Broad Network Access: System Center 2012 SP1 Configuration Manager and Windows InTune interoperate for Mobile Device Management (MDM) and universal application distribution-supporting Android and Apple IOS devices alongside Windows PCs and Windows Phones. Microsoft identity management is open and includes Google, Yahoo, Microsoft ID, and Active Directory authentication providers.
3. Resource pooling: VMM pools data center fabric for VM provisioning, Operations Manager pools management servers and gateways. Windows Server 2012 Storage Spaces pools disks of any type and from any disk controller.
4. Rapid Elasticity: Provision or delete a VM in minutes. Add disks to highly available storage pools without adding array controllers. Extend on-premise network and compute resources to clouds for peak and burst capacity.
5. Measured Service: VMM with Operations Manager produces charge-back reports on a per-cloud/per-service basis that let you quantify exactly how much a service is costing.

Microsoft’s secret weapon for renewed industry relevance

Once you become familiar with the features and capabilities of Windows Server 2012, System Center 2012 SP1, and Windows Azure, you can see a complete canvas emerge–upon which you can draw a low-risk, long-term, and high-yield strategic IT roadmap for an organization. CloudOS is a compelling concept because it has the appeal of predictability and positive ROI that are Microsoft strengths in the enterprise computing space.

That Microsoft is building a great product here can be seen in Microsoft’s Server & Tools division being the company’s fastest growing division Q4 last year. Internally Microsoft knows they are onto a good thing, late in 2012, they reorganized to merge the Windows Server and System Center technical teams. Breaking down organizational barriers between the server and the tools teams makes a lot of sense when you honestly want to blend the OS and the tools into a single entity–CloudOS.

Read More

The myth of the always-on cloud

Takeaway:  Advises caution when considering uptime claims for cloud services and software. Realistic expectations will lead to a better experience when moving to the cloud.

One of the big promises of cloud computing is the idea of always-on. The cloud as a whole - infrastructure, platform and software - is supposed to be available at all times. Service providers from every layer give 99% and above availability guarantees and everyone claims that their services are resilient and failure-tolerant. While the track record of different providers can vary wildly, the real problem is that clients often forget that even 99.99% availability does not mean that a service will be accessible all of the time.

Let’s disregard for a moment the large differences between what each provider describes and considers as being available, and look exclusively at the numbers. 99.99% uptime of a service over the course of a year means that the service can be offline for about 52.56 minutes per year, or roughly one minute per week. This could account, for instance, for a server reboot every other week or so. As the uptime guarantee decreases, the downtime numbers obviously grow: for 99.9% uptime, the service can be offline for 525.6 minutes, or 8.76 hours, in a year; for 99% uptime, it would be 5256 minutes, or 87.6 hours, which is more than an hour and a half every week. While some of these numbers may seem small, they can severely impact systems and processes that aren’t ready for them.

Handling failures

The first thing that anyone who is looking to use cloud-based services or any kind must consider is the possibility of failure: what happens when I try to call the service and get back an error response or, even worse, get no response back? Retrying a request is an obvious answer, but also a problematic one. If a service goes offline for a significant amount of time, a “retry loop” can trap an application or create unexpected situations from which it can’t recover.

Even worse, issuing many retries can create a bottleneck at the receiving service, with even worse consequences. An interesting example of this was last year’s October outage of AWS, where they stopped accepting requests for creating EBS volumes and EC2 instances due to an excessive number of errors. In this sense, retrying requests can create a cascade of failures in interdependent services that becomes harder and harder to recover from.

Handling failures means having contingencies in place to handle unforeseen and unexpected situations. This not only means reacting and dealing with obvious failures, but also with situations that don’t clearly represent a failure. Let’s take a system that automatically launches virtual machine instances to do some processing: it must be prepared to handle the situation where the request for a new virtual machine is denied, but also for the situation where it receives a normal response for the request, but the virtual machine is never launched.

True availability

Availability issues become even more pronounced when we consider the track record of service providers. While almost everyone will promise aggressive SLAs (99%+ uptime), the fact is that many of the top tier providers routinely fail to deliver the promised levels of availability. If the systems that make use of these providers aren’t ready to handle a provider being unavailable for long periods of time, they will fail spectacularly in real life.

Another important point to take into consideration is that a system can only be as available as its underlying components. Many cloud software-as-a-service providers offer uptime guarantees that they can’t hope to match in real life, because these guarantees surpass what they are getting from their own infrastructure providers. If you’re looking for cloud-based software, always beware excessive promises. At the same time, take into consideration what it means to only have 99% availability: if the software were to stop working, can your business survive?

None of the issues discussed here are new. Most of them have been around since the advent of client-server architectures some decades ago, but sometimes users and developers forget the lessons of the past just because they are dealing with new technology. By remembering that cloud-based services are just like any other IT system and that eventual failures are expected, we can avoid many headaches when moving to the cloud.

Read More

The future of data centers: Is 100% cloud possible?

By Guest Contributor

Takeaway: Guest blogger Robert Offley explains how the market is shifting today, what barriers remain for total cloud adoption, and if an evolution to 100% cloud is likely to occur.

Today, many midsize and large companies are taking a piecemeal approach to cloud computing. While the majority of companies are experimenting with cloud computing, few have migrated their entire infrastructure to a public or private cloud environment. A survey published in January by Enterprise Strategy Group of more than 600 enterprise and mid-market companies globally found that 27% were using public cloud IaaS services, up 10% from a similar survey published in early 2011. 24% of respondents said they haven’t pulled the trigger on a cloud solution yet, but plan to at some point in 2012.

While cloud offerings from behemoths like Amazon are at the forefront of press coverage, many companies are seeing compelling incentives to gradually transition their infrastructure to the cloud, and construct more customized and hybrid solutions often offered by smaller and lesser known players in the industry. A hybrid approach enables organizations to ease the transition to cloud, and alleviate any risks associated with a complete cloud migration. What does the role of cloud computing look like for organizations in the near future?

Cloud computing is becoming ubiquitous in today’s businesses, with the rising popularity of free SaaS applications like Dropbox and Google Apps, among others. For IT decision makers, however, the decision to migrate all back-end and mission-critical applications to the cloud may not be the most appropriate - at least for now. In many instances, a gradual progression to the cloud is an ideal strategy. This is why the Hybrid Hosting model is an attractive option for many businesses. Hybrid hosting combines aspects of physical hosting and cloud computing, and facilitates a seamless progression to a completely virtual hosting solution. Businesses are able to pick and choose what services are required, the types of management levels provided by the vendor, and whether or not their data and applications are stored on dedicated hardware or shared virtual servers.

Taking the Hybrid approach is also a necessity for some organizations, since there are still some significant barriers to cloud adoption:

Security: High-Risk applications, such as those that store financial, medical, or other sensitive information, may not be suitable for a public cloud environment. It is also important to verify the level of encryption included in a public cloud offering. Furthermore, organizations must determine exactly who has access to the data stored on virtual machines through the administration panel. Security concerns must be evaluated on a case-by-case basis. Once this occurs, then the cloud can be evaluated as an appropriate or inappropriate solution.

Compatibility: Many companies are still using legacy systems that predate virtualization technologies. As a result, these systems will not work well in the cloud unless they are significantly reconstructed. Additionally, end-user licensing agreements may not allow for deploying these systems virtually. Legacy systems are typically not appropriate for the cloud due to issues surrounding compliance, end-user support, licensing, high-speed data access, and internal skillsets. The process of transitioning a legacy system to be cloud-ready requires a significant investment in time, resources, and money.

Geography: One downfall inherent with public cloud offerings is the physical location of data. Many organizations require confirmation of where their data resides geographically. Unfortunately, many public cloud offerings disperse their infrastructure throughout multiple jurisdictions. As a consequence of this reality, customer data can become susceptible to different legislations covering specific geographic boundaries, such as the US Patriot Act, PIPEDA (Personal Information Protection & Electronic Documents Act) in Canada, and equivalent legislations and guidelines in the EU and many countries around the world. For example, with Canadian-owned and operated organizations looking to store their data in the cloud, the risk of having their data being susceptible to international legislation has proven to be a deterrent to adopting cloud hosting solutions.


Is the future of data centers 100% cloud? Once more organizations become comfortable with cloud computing, and after the barriers listed above are remedied, the progression towards 100% cloud adoption will increase. The Hybrid hosting model will remain integral to many organizations wishing to transition all of their applications to the cloud. With that being said, an evolution to 100% cloud is not likely to occur anytime soon. Due to compliance, compatibility, security, geographic concerns, and internal skillsets, many organizations will still require physical hosting services and on-premise infrastructure to fulfill their requirements well into the foreseeable future.

Robert Offley is the CEO of CentriLogic, a provider of managed hosting and cloud computing solutions.

Read More

The 10 traits of the smart cloud

By Thoran Rodrigues

Takeaway: Thoran Rodrigues interviews Dr. Satwant Kaur, a leading expert on emerging technologies, about the required traits of the “smart cloud.”

I’ve recently had the opportunity to have a conversation with Dr. Satwant Kaur on the topic of Smart Clouds. Dr. Kaur has an extensive history in IT, being the author of Intel’s Transitioning Embedded Systems to Intelligent Environments. Her professional background, which includes four patents while at Intel & CA, 20 distinguished awards, ten keynote conference speeches at IEEE, and over 50 papers and publications, has earned her the nickname, “The First Lady of Emerging Technologies.” As such, she is now putting forth the idea of Smart Clouds, and I got to talk to her a little bit about this idea and its implications for the cloud computing industry and for businesses in general.

Q: What exactly are smart clouds, and what is it that makes a cloud smart?

A: In the emerging computing model, computers are mobile, devices are connected, applications are interactive, development is collaborative, and service delivery needs to be immediate. Cloud computing can enable all these capabilities for businesses while optimizing costs and resource usage, since infrastructure (IaaS), platforms (PaaS) and software applications (SaaS) can be delivered on a usage-based service model. Users can get access to what the want as and when they need, instead of having to pay for any idle computing resources.

While this is excellent in theory, there are still some major gaps in the capabilities of cloud providers today that act as deterrents to businesses that are looking to make the move to the cloud. The Smart Cloud is a framework of the necessary traits or characteristics that any cloud offering (regardless of it being IaaS, PaaS or SaaS) should have in order to fully answer the needs of business users everywhere. The ten most important traits are as follows:

1. Smart Clouds can be built modular by using Cloud-In-A-Box cells.
2. Smart Cloud services are secure.
3. Smart Clouds provide Automated Management of Cloud Services.
4. Smart Clouds perform to meet needs of the real-time nature of business.
5. Smart Clouds are available for business-critical services.
6. Smart Clouds enable businesses with governance and control.
7. Smart Clouds are provisioned rapidly by  self-service.
8. Smart Clouds have solutions to manage the cloud environment.
9. Smart Clouds have solutions to transform the infrastructure.
10. Smart Clouds have integrated social collaboration solutions.

Q: Can you explain in a little detail each one of these traits?

A: The first trait, about modularity and the “Cloud-in-a-box” model, relates mainly to the lower levels of the cloud stack, especially infrastructure-as-a-service. It basically means that any cloud offering should not only be fully redundant, but also easily scalable. A smart cloud provider should be able to plug-in computing/storage/networking units as needed to increase capacity without any impact to existing systems. At the same time, each individual unit should be completely redundant and self-reliant, ensuring that there is no single point of failure for the entire system. This translates not only to better scalability, but also into better resiliency, since, if one unit fails, the rest would keep working as usual.

When we discuss security, we are focusing not only on technological security through encryption and access control, but also on procedural security. A smart cloud should perform automated security testing (penetration testing) on itself at every layer, from the hardware to the operating system to the applications that run on top of it. Furthermore, it should simplify security management, so that privacy protection policies and security policies can be easily deployed across the whole system.

Traits 3, 7 and 8 go together hand-in-hand. They relate to end users having the ability to dynamically expand their resources as needed without having to go through complex interactions with their service providers. The smart cloud should offer something as close to a “one-click purchase” experience for cloud resources as possible. At the same time, the user needs effective management tools that allow them to quickly see what is going on at every level of their cloud, regardless of the mixing of private and public resources. Finally, the smart cloud should also offer alternatives to automate management, so that resources can be brought on-line as the system itself detects a need for them, or, conversely, that resources can be shut down when they are no longer needed.

Traits 4, 5 and 6 are directly related to business needs. To address the real-time needs of businesses, the smart cloud needs to constantly monitor itself not only from a back-end point of view, but also from an end-user point of view, so that IT always has an up-to-date view of what is going on at both ends of a service, and knows exactly what kind of experience the end-user has today. With regards to availability, a smart cloud service should offer not only the expected SLA, but also full transparency into service status and uptime, with the tools users need to manage it. Finally, the smart cloud should offer all the tools businesses need to deploy their own governance rules, to audit any system as needed, and to ensure that specified policies are properly applied.

Trait 9, about infrastructure transformation, implies that smart clouds should bridge the gap between private and public resources. They should enable companies to migrate their internal IT into a service-based model, and also allow them to easily interoperate with other services, be they public or private. This would bring about an easy path of migration for businesses so that they can exploit cloud benefits.

Finally, the last trait talks about social collaboration. Many providers today employ communications solutions that include social media to give status reports, but these tools are mostly about broadcasting, and not collaboration. Social collaboration is a two-way street. A smart cloud solution enables users not only to collaborate and interact with other users, but also with the cloud itself.

Q: Okay, so most of these actually sound familiar. Are the clouds today already smart?

A: Well, I would say that cloud vendors today fulfill anywhere from 20 to 80% of the traits outlined above. The top vendors today do offer a lot of these capabilities, but there are still issues that plague cloud implementations everywhere, and these are the hardest ones to address.

Q: So what are these issues? What are the main issues for the widespread deployment of smart clouds?

A: Interoperability is probably the main issue today. Today there are too many platforms, too many technologies, too many domains of expertise, and too many vendors to coordinate and manage. A smart cloud therefore needs to offer a path to interoperability between services, as well as service brokers that can make it easier for users to get at the services they need when they need them, without having to go through a complex hiring process. In a sense, this is the hardest step to overcome, since interoperability may not be in the best interest of all vendors.

From a technological perspective, we still need to see some developments related to service standardization (which would improve interoperability) as well as improvements in service management and delivery.

Q: And how about security? Whenever I talk with businesses, they express a clear concern with the security of cloud offerings.

A: The main obstacles in making the cloud secure are how to deal with federated identity, authorization and entitlement and how to integrate with the existing security infrastructure that businesses have already deployed. At the same time, smart clouds need to have automated intrusion detection and prevention, as well as powerful tools for audits and compliance reporting.

Q: Finally, what are the business benefits from the smart cloud?

A: The smart cloud should be seen as a framework to be pursued by businesses when looking to the cloud. No matter if they are building their own private cloud or if they are going with cloud providers, they should try to satisfy as many of the traits defined above as possible. Those traits that aren’t fulfilled from the start should be clearly outlined on the roadmap for the services being hired.

That said, smart cloud services can enable businesses to dynamically right-size their IT consumption, optimizing costs and improving availability and scalability. At the same time, it would allow businesses to safely break down barriers between them and their consumers and even other businesses, improving collaboration and simplifying access to information. Finally, smart cloud services improve the interoperability of business processes and systems, resulting in a more resilient and agile business.

The issue with traditional clouds was that it did not meet the complete needs of businesses. The Smart Cloud Technologies are emerging, and hence businesses are just beginning to adopt. But this is a quickly growing area.

Read More

2012 AWS Cloud Summit: An intro to Amazon's cloud tech for businesses

By Nick Hardiman

April 27, 2012, 12:35 PM PDT

Takeaway: Nick Hardiman attended the 2012 AWS Summit in London and shares his impressions of the event and what he learned about Amazon’s cloud technology. Also, there was free beer.

Amazon and its business partners produce a continual stream of summits, seminars, webinars and other events promoting the use and understanding of Amazon Web Services. These events are scattered all around the world, in cities such as NYC, Chennai and Seoul.

An AWS traveling theatre arrived in London recently so I headed down there to hear what they had to say. Would the event be an Amazon infomercial, distributed computing theory, or industry news on the state of the cloud?

The venue was the Methodist Central Hall in Westminster, London. It’s just around the corner from Big Ben and the Parliament of the United Kingdom, the heart of the British government. This is the area where Victorian captains of industry ruled the British Empire. Now the new captains of industry are just passing through on their world tour (next stop: Melbourne).

The AWS London event was actually two separate events on two consecutive days, aimed at two types of customer.

• Event 1: AWS Cloud for Start-Ups & Developers, where Amazon architects demonstrated solutions to the nerds.
• Event 2: Cloud Computing for the Enterprise, where Amazon chiefs powerpointed the suits.

At the beginning of each event, the events staff in the Central Hall dished out name badges, pastries and drinks in the reception hall to a crowd of a couple hundred people. A few AWS partner booths are on the borders of the room, but it’s a token effort - this is no big exhibition with a footfall of thousands.

Event 1: AWS Cloud for start-ups & developers

The crowd for part 1 was more “heavy tech” than “business admin”. There was a touch of the buzz of a start-up networking event where people swap stories and practice their pitches, but it was mostly a collection of men standing around looking awkward, perhaps because they were separated from their development environments. It was a relief when the technical talks started.

In a side conference room one of the Amazon marketing managers introduced a succession of speakers. Amazon solutions architects illustrated AWS cloud services with live demos, which actually worked properly.

Happy AWS customers described how they used the AWS cloud in their business. Well, partly. Ian van Reenen, CTO of remote management company Centrastage, gave a fair description of both his company and their use of AWS, such as how dynamoDB performs for them during their traffic spikes.  Anthony Rose, CTO of TV-related app company Zeebox, painted a picture of his company’s social media awesomeness, with a few AWS facts stuck on the end for decoration.

The event finished off with two hours of networking and free beer back in the reception room. Amazon CTO Dr. Werner Vogels was around at beer time, attracting smart-dressed guys with wheeled suitcases into his vortex.

Event 2: Cloud computing for the enterprise

This enterprise event had a different feel to the previous start-up event. Delegates were employees sent to learn more about Amazon. The average age went up by a decade. Casual clothes were replaced by business suits.

The day’s agenda was designed to equip an enterprise manager with the tools to make better use of AWS. Each speaker used a slide show for illustration, not the live demo favored by developers. The talks were delivered by VPs, not solution architects.

The keynote speaker was Dr. Vogels himself, giving a superb troop rally on the ways Amazon is transforming business.  He used examples of vast cost savings, threw in similes - did you know Amazon is like the Kit car in Knight Rider, and a firewall is like the moat around a castle? - and enthused about many Amazon services.

Real-life examples from the day were all enterprise scale: how Samsung saved millions of dollars, how Animoto scaled from 50 to 5,000 servers in 3 days, and how many international standards AWS complies with. The happy AWS customer talk by Robin Meehan, CTO of the system integrator Smart421, was also enterprise-scale: he described how they created a DR solution for Haven Power.

Was it worth going?

So what did I get from all this free food and speeches by big players?

Background knowledge. I gained a clearer understanding of who uses AWS spot instances. I felt the excitement of some enthusiastic entrepreneurs. I got the point of some AWS business partners. It’s not information I can use tomorrow. It’s illumination of the cloud territory we find ourselves in.

Read More

Stress-testing a web service: The basics

By Nick Hardiman

April 24, 2012, 10:34 AM PDT

Takeaway: Nick Hardiman describes his preparations to stress-test his simple Drupal-based web service built on an Amazon EC2 machine.

I built a Drupal installation on Amazon EC2, and now I want to find out how scalable my new Drupal service is. To do this, I am going to run my first set of load tests (AKA stress tests), to push my service to the limit.
When more requests arrive at my website, how does it behave? Does it handle the increased load without affecting the performance? Does it remain reliable?
Right now, I don’t know. I cannot offer a fully operational cloud service without knowing how it scales, so I need to kick off a few tests.
I use a few free tools to torture the homepage of my new customer service, rather than using a commercial service like Loadstorm. I put my service under increasingly heavy loads to see what happens and measure the results.
Later, I will need to fix the problems I find.

This service is born to lose

One small Amazon EC2 machine is weak. A small physical machine may handle a reasonable web workload, but a small EC2 machine won’t. These tests illustrate why you need to be able to scale up (use a bigger machine) and scale out (use many machines).
I know my service will perform poorly. It’s so poor it can’t even afford to pay attention. This is how I ensure awful service.

• I use a small EC2 machine type. Apache VMs come in different sizes, from micro to massive. A small VM has few resources, which makes it a tight fit for a web service.
• I only use one EC2 machine. I have two identical VMs to share the work, but I run the tests on one machine only.
• MySQL has not been tuned for Drupal. I’ve made no changes to buffer size, not hunted for slow queries, and run no engine checks.
• I don’t use a cache. Caching using an application like memcached or varnish is a popular way of increasing speed. Just turning on the Drupal cache speeds up response many times.

Figure A

My load testing toolkit

I could use a third-party service to run more comprehensive tests with web interfaces, such as the Jmeter cloud and SOASTA CloudTest but I don’t really need anything that clever for now.
I generate and monitor the extra load using a few applications. These are all command line tools, so they are not very intuitive.

• top (top process statistics). This is a process monitor that shows me what is happening to the system.
• vmstat (virtual memory statistics). This is like top, but displays information in a different way.
• ab (Apache HTTP server benchmarking tool). This is a web site load generator. I use ab to give my service an increasingly hard time.

top (top process statistics)

The top command is one of the ten most useful Linux commands. Top displays information about processes and what they are doing to the system.
I want to get a baseline by using top while the system is idle, before I run the load tests.

root@ip-1-2-3-4:~# top

top - 17:12:40 up 15 days,  3:33,  2 users,  load average: 0.46, 0.69, 0.34

Tasks:  71 total,   1 running,  70 sleeping,   0 stopped,   0 zombie

Cpu(s):  0.0%us,  0.0%sy,  0.0%ni,100.0%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st

Mem:   1737564k total,  1117352k used,   620212k free,   168248k buffers

Swap:  3020212k total,        0k used,  3020212k free,   518672k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND

    1 root      20   0  8356  800  672 S  0.0  0.0   0:12.68 init

    2 root      20   0     0    0    0 S  0.0  0.0   0:00.00 kthreadd

    3 root      RT   0     0    0    0 S  0.0  0.0   0:00.00 migration/0

    4 root      20   0     0    0    0 S  0.0  0.0   0:01.08 ksoftirqd/0

…

Understanding the numbers is difficult. The top command fills up my command line interface with a lot of data packed into two halves.

• The top half - about six rows - is a dense display of information about the state of the system, such as how busy it’s been, memory used and uptime.
• The bottom half is a list of the top processes. They are ordered by how much CPU they use, with the biggest CPU hog first.

The procedure for using top is pretty straightforward.

1. Open a CLI.
2. Run the top command. A display like the one above appears.
3. Watch the numbers. Every few seconds a few of the numbers change.
4. When you’ve had enough, type the letter q to quit. The command prompt appears.
5. Close the CLI.

There is a lot of information here: it is compressed to pack a lot into a small space. The more you use top the more numbers you can understand. It’s a bit like staring at a stereogram until a 3D picture appears.

My first measurements

Even before I run my first load test, I can make some useful observations about my EC2 machine.
In the example above I see an idle system. The CPU is 100% idle and no swap space is being used. It’s pretty obvious to a system administrator that this EC2 machine is doing nothing.
The load average is 0.46. The load average is an estimate of how much the box is doing compared to what it can handle - 1 is roughly 1 CPU working flat out, but keeping up with its workload.
Strangely, this idle box is putting in about half a server’s worth of effort. Shouldn’t a box doing nothing have a load average of zero? What’s happening is the hypervisor is stealing my VM’s capacity and giving it to other busier (and maybe higher-paying) customers. It’s the same theory that an airline uses when it over-sells seats on a plane, relying on some passengers to not show up.

 

Translate in indonesia 

Read More