The Frugal Networker

Frugal Networker podcast with Palador co-founder Benjamin Robbins. We discuss mobile device management for your company. We cover the risks, the advantages and some of the myths surrounding BYOD. Follow Benjamin Robbins on Twitter @paladorbenjamin or catch his Remotely Mobile blog. 21 minutes. MP3 format. Rated G for all audiences.

Palador_Mobile_Device_Management_and_BYOD

Podcast with Jason Perlow, Senior Editor and Blogger at ZDNet. We discuss Data Centers past, present and future with our perspective on what data centers will look like in the future. We also cover how data centers of the future might become the next utility and how differently you’ll use computing services.

MP3 format. 26 minutes. Rated G for all audiences and venues.

jason_perlow_datacenter_discussion_06072012-e

This post was written as part of the IBM for Midsize Business program, which provides midsize businesses with the tools, expertise and solutions they need to become engines of a smarter planet.

Red Hat, Inc. is the clear leader in the Linux market. It has the strongest, commercially supported Linux distribution and the best-performing virtualization solution for servers and desktops. It is the first billion dollar open source company in the world and is the most successful Linux company thanks in part to its dedication to the open source community and free software. Red Hat Enterprise Linux (RHEL), its flagship Linux product, is the one to watch in the data center for enterprise-level workloads including databases, application delivery and virtualization.

In addition to high performance standalone server computing and virtualization, KVM offers competitive technology for virtual desktop computing. Virtual desktop computing or VDI moves your desktop operating system away from local hardware and places it on enterprise-level server systems in your data center.

What you’re looking for in a hardware/software combined server solution is performance, upgradability, support, scalability and affordability. Red Hat and IBM have formed an alliance to make this perfect recipe a reality with Red Hat Enterprise Linux and IBM Server x series server hardware.

Red Hat Performance and Lower TCO

You can hardly deny Red Hat’s superior performance when Oracle, owner of Solaris, develops and runs its own Linux version (Oracle Linux) based on Red Hat Enterprise Linux. Oracle continues to certify its database product on Red Hat Enterprise Linux (Oracle certified its 11gR2 database on RHEL 6 on March 22, 2012).

Red Hat continues to make marked improvements in the following areas for its RHEL product line:

• CPU/Kernel – Non-Uniform Memory Access (NUMA), scheduling, Read-Copy-Update (RCU) and extreme guest virtual CPU scaling.
• Memory – Transparent Huge Pages for optimal hardware-based virtualization.
• Networking – VhostNet – Network stack moved to kernel.
• Block – Asynchronous I/O, MSI caching protocol, vectored I/O.

These ongoing enhancements make Red Hat’s Linux and its associated KVM virtualization platform an efficient combination for standalone or virtualized systems.

For specific operating systems, you’ll experience near native performance for virtual machines at all operational levels. Red Hat recommends that you distinctly identify your operating system at installation so that OS-specific optimizations can be applied. For Linux operating systems, the boost in guest NFS write performance is more than 12%–a result of specifying the OS type at VM creation.

But, performance in the operating system isn’t useful without the underlying hardware to support it. RHEL 6’s design takes advantage of hardware enhancements such as NUMA, hardware-assisted virtualization, networking and block device hooks that go unused if the hardware’s design is flawed or doesn’t contain supported features. In other words, you can’t architect a solution based solely on software or on hardware alone—you have to consider all aspects of a solution.

IBM’s System x servers, for example, are the perfect mix with RHEL 6.x to lower your TCO. With this combination, you can expect as high as a 20:1 server consolidation ratio and up to 95% lower power consumption with System x series hardware.

To further save money, you can virtualize workloads on KVM that were once thought to be standalone server only capable. Some examples are IBM’s DB2 database, Lotus Domino, Tivoli products and Websphere.

Built-in Scalability

Many people talk about scalability, and it’s a cute buzzword to toss around, but RHEL can actually do it. Scale, that is. Scalability is important, if you need it. If you don’t need it, then it’s really a non-issue. As it stands now, RHEL 6.x’s scalability far exceeds the practical limits of current mid-range (x86_64) hardware. However, if you require filesystems that support very large file sizes (up to 16TB), 64 processors or multiple terabytes of memory, RHEL is at home on a variety of architectures and hardware types—from x86, AMD64, Intel 64 on the low and mid-range end to IBM Mainframes at the high end. The scalability is available when you need it.

One of the problems with scalability is that you can scale yourself into a performance conundrum by relying on theoretical limits instead of the accepted practical limits. For example, RHEL 6 has a supported 2TB memory limit and a 64TB theoretical limit (x86_64 architecture), so the edict comes down from above that the system administrator should double a system’s memory from 64GB to 128GB to increase performance. Everything works in theory but without some ‘tweaking,’ the system administrator finds that performance actually decreases in this scenario. Unexpected results, if you don’t understand capacity management*.

Desktop Virtualization

If you want to invest in virtual desktop infrastructure (VDI), you should seriously consider KVM for its extreme performance delivery. KVM supports both Windows and Linux desktops with exceptional end user experience via its (now open source) SPICE dynamically adaptive remote rendering protocol. SPICE delivers a desktop experience that is virtually indistinguishable from a local desktop experience complete with multimedia support.

VDI can be another TCO-lowering move for mid-sized businesses. With it, you can move all of your operating system maintenance to the data center so that your users can focus on productivity instead of dealing with antivirus updates, application updates, Windows updates, reboots and interruptions by support staff to install an application or update. Visits to staff desks will be limited to hardware replacement and deployment, which takes minutes instead of hours since you’ve removed the operating system from the mix.

VDI requires some initial investment but the long-term savings is worth it, if your organization has the capability of moving that direction. An assessment by a competent IT strategist will help you decide if VDI is right for you.

IBM and Red Hat

IBM was an early adopter of Linux and formed partnerships with Red Hat that are now well into their second decade. IBM also defended Linux against the lengthy and expensive lawsuit launched by SCO. IBM’s support of open source and Linux is long-standing and proven by its release of patents to the open source community to promote innovation and industry growth in an open and collaborative atmosphere.

*Rely on Red Hat’s and IBM’s experience, when attempting to boost performance output.

This post was written as part of the IBM for Midsize Business program, which provides midsize businesses with the tools, expertise and solutions they need to become engines of a smarter planet.

To engage in a server consolidation project, means that you have something in mind for your business through cost savings, delivery improvement or business realignment. But, regardless of the reason, server consolidation has the side effect of lowering costs and that, in itself, is reason enough to expend the energy for such an undertaking. What about incorporating a virtualization project into the overall picture? Server consolidation is, by design and by definition, a money-saving undertaking, so why consider adding cost back into your data center with virtualization. The answer, of course, is cost savings.

The old business adage, “You have to spend money to make money,” holds true for virtualization. Virtualization will cost you money. You have to pay for host hardware, software licenses, training, network setup and possibly reworking power to the racks (if you switch to blade servers). It’s hard to see how such a large financial commitment makes financial sense but it does in terms of total cost of ownership (TCO). Consolidation and virtualization are the means to the end, which is that of cost-savings.

Exploring Server Consolidation

Server consolidation is a labor and time intensive project for you and your staff. It takes time to assess which systems have the capacity to share workloads and which ones are idle. You’ll find that the process leaves you with spare systems or ones that can be redeployed. You should plan to decommission systems that are at or near their end of life (EOL) date. Additionally, you should consider decommissioning or repurposing systems that you find to be underutilized.

Once the numbers are in on your consolidation efforts, you can then turn your attention toward moving to a virtualized infrastructure from a purely physical one. You should also consider moving toward some cloud-based services to better leverage your computing and labor resources. Virtualization coupled with cloud computing services and storage creates an “always on” environment for you and your customers. Cloud computing might also have the unexpected effect of lowering your computing overhead costs by allowing you to outsource services and labor to external providers.

Assessing Performance

Before you can do any real consolidation work, you have to gather some empirical data on your systems’ performance. Don’t take this phase of the project lightly or rush it. You need to know utilization data for each system in your inventory that’s included in the project. Focus on systems whose average utilization is below 40 percent for a first pass. Idle systems, or those that are mostly idle, are prime candidates for consolidation. Secondarily, turn your attention toward systems whose hardware is overworked. Moving workloads works both ways in a consolidation effort.

Decommissioning Hardware

Hardware within six months of EOL should be marked for decommissioning. Any other hardware not suitable as a virtual machine host or other standalone system workload (Domain controller, database server, etc.) should also join the decommission list. One of the primary purposes of a server consolidation and virtualization exercises is to minimize the number of physical systems in an environment to virtual machine hosts and a few physical workload systems.

Migrating to a Virtual Infrastructure

Moving your systems from a physical state to a virtual one is easy. For physical systems that must remain ‘as is,’ use a Physical to Virtual (P2V) conversion tool such as VMware’s Converter, Microsoft’s System center Virtual Machine Manager, PlateSpin Migrate, Quest’s vConverter or Citrix’s XenConvert. Whatever tool you use, you’ll probably want to use one that features live migration, which means that you can convert a live (running) system to a virtual machine without interruption.

Your other conversion option is to install a fresh virtual machine and setup its applications, users, domain membership and networking, while its physical counterpart is still in production. Duplicate the physical system in virtual form and then, just before switchover, copy all data to the new system, change the IP address, change the system name and then reboot. The length of the process depends on how much data you have to copy. It is preferable to keep both machines live and available until the virtual machine cutover has been verified as production ready. The physical system will have to be renamed and setup with a new IP address to maintain network integrity during the testing phase.

Considering Cloud-based Services

One of the biggest surprises to any virtualization effort is storage. It’s shocking to realize how much storage you require for your virtual infrastructure. For this reason, it’s wise to seriously investigate cloud-based storage services for your virtual infrastructure. It’s easier but often less cost effective to use private cloud storage. If you choose not to use cloud-based storage for your primary storage needs, then you should certainly entertain the use of it for disaster recovery (DR) and archival purposes.

Cloud-based services can also include application hosting or adding additional bandwidth on demand for your computing environment. For example, you can create additional virtual machines on a cloud provider site at a very low cost. Keep those virtual systems powered off until you need the extra bandwidth and only use them during peak periods, such as those high traffic times associated with promotional events. Cloud services are an excellent and cost-effective way to extend your reach at a very low comparative cost.

Reaping the Rewards

Once you’ve created your virtual infrastructure for the server consolidation move, your cost accounting job begins. Consider that a lot of the work performed by multiple groups shifts to your virtual infrastructure administrators, who typically are system administrators. This group handles the host system operations, the virtual network creation, virtual switch creation, VLAN setup and virtual machine maintenance.

Having a virtualized infrastructure also means that you’ll need fewer people managing the environment, since it’s now self-contained in its own virtual realm. You’ll need a SAN or NAS storage team, a network team and a system administrator team. Of course, your database administrators, developers and applications support won’t change but you should be able to operate with fewer primary support staff.

The principal reason for the reduction in primary support staff members is that you have only a few physical systems. For example, rebooting a physical system is risky because it can hang on a hardware error or require a ‘reseat’ in the case of a troubled blade system. There are no such problems when dealing with virtual machines. When a virtual machine hangs on boot, several actions that a system administrator may take require no direct physical intervention. It is also this location independent quality of virtualization that allows companies to outsource support to third-party companies at a lower cost compared to that of in-house employees.

Notes for Mid-sized Businesses

Don’t be intimidated by the thought of moving to a virtual infrastructure, whether you’re at the low end of mid-sized or close to the high end, virtualization has something to offer you. Consolidation and virtualization can take that overfilled server room and turn it into an efficient workspace for your commercial and internal applications. It means business agility for you so that you can respond to changing customer needs and to a business climate that is always in flux. Virtualization means that your business is also more mobile than ever before.

Should you decide to move to a cloud-based infrastructure and away from self-hosted applications, you can move your virtual machines to a cloud provider without losing any productivity or business continuity. You’re no longer tied to a single location or to a single data center. Your consolidation and virtualization efforts pay off in many ways: mobility, agility, scalability and frugality.

Summary

For many IT shops and companies, server consolidation and virtualization are different steps in the same process, although they don’t have to be. Each process can stand on its own. Server consolidation projects are common in data centers, when hardware inventories discover system sprawl or when clever administrators uncover idle systems. Virtualization activities often revolve around the desire to save money by decreasing the number of physical systems taking up rack space.

Both activities will save you money but server consolidation has a more immediate and dramatic effect on the bottom line. Virtualization can cost a lot of money but is a longer-term investment. The best advice is to request proposals from vendors and look at the numbers with the knowledge that virtualization reduces your TCO in spite of the up-front costs. Remember that fewer physical systems mean lower costs, fewer required staff members and less direct interaction with the computing environment. The time, effort and money you invest in server consolidation and virtualization pay off in ultimate savings for your company.

This post was written as part of the IBM for Midsize Business program, which provides midsize businesses with the tools, expertise and solutions they need to become engines of a smarter planet.

Web Application Servers are nothing new in the tech world but many business managers, application developers and systems administrators still don’t understand what they are or why they’re needed. The Three-Tier Architecture allows architects and developers to create a dynamic and relatively secure method of delivery dynamic content to users. Web Application Servers are the key component in this three-tiered delivery model. A Web Application Server (WAS) not only delivers dynamic content but it also contains the business logic, the business rules, the data access and a modulated connectivity path between the data and the data consumer or user.

Three-Tier Architecture (3TA) is the design that results from splitting individual services onto multiple systems and into multiple layers—both physical and logical.

Logical vs. Physical Architecture

3TA consists of three distinct tiers or layers: Presentation, Application and Data. When speaking of 3TA, most discussions refer to the logical architectural layout. Logically, the Presentation Tier consists of client computers and web services that provide the user interface. The middle or Application Tier contains the business logic, the rules for information processing and the data access components. The Data Tier contains the data and data storage.

The Presentation Tier or layer deals with user interaction and user experience. This layer transmits requests from the user and presents the responses back to the user in a readable format. The Application or middle Tier receives requests and either responds directly back to the user or queries the datastore and configures a response for the user.

The third tier is the Data Tier and its purpose is to store data and to provide that data via requests from the Application Tier. The Data Tier never comes in direct contact with the Presentation Tier.

Three-Tier Design Advantages and Disadvantages

Advantages

• Scalable Design – The addition of new servers and load balancing can grow an environment to accommodate large numbers of client connections.
• Parallel Development – Developers and DBAs can work simultaneously and independently on the different layers (tiers).
• Superior Performance – Separation of CPU-intensive, memory-intensive and I/O-intensive operations increases and extends performance of all components.
• Increased Security – Physical and logical separation of components can increase security.
• Improved Availability – Redundant server members decrease the severity of outages.

Disadvantages

• Design Complexity – Multi-tier architecture is more difficult to implement than single tier.
• Increased Maintenance – Designated systems (Web, Application and Database) often have their own maintenance schedules and windows that might prove cumbersome to production.

Physically, the servers have separation from one another as well. Client systems are part of the Presentation Tier and remotely located (physically separated) from the other tiers. To further separate the Presentation Tier from the Application Tier, architects place web servers in a DMZ so they’re network connectivity faces the Internet on one side and the corporate LAN on the other. On the LAN side, a firewall limits the TCP/IP connectivity to a few destinations: The Application Servers. This limited connectivity reduces the attack surface for would-be intruders.

Connectivity constraints between physical tiers continue from the Application Tier to the Data Tier. Architects further isolate database systems and data storage by only allowing database access from the application servers in the Application Tier. And, only the database systems directly access data storage.

This isolation of tiers is not only a security measure but also one of performance and one of availability. By imposing limits on the number of connection origins, from the web servers to the application servers and from application servers to the database servers, the potential for capacity overload is very low. Additionally, by spreading the load over multiple web servers, multiple application servers and even multiple database instances via load balancing mitigates performance problems due to high traffic bursts.

For availability, multiple systems provide a resource pool that creates a cushion against service outage in case of a single system’s failure. Administrators will remove the failed system from load balancing until it’s replaced or repaired.

Application Server Role

The role of the application server or the WAS is to receive requests for dynamic data from web servers, to filter and to shuttle those requests to the database, to gather and to organize the requested data and to deliver it back to the user. The application server also performs security checks including verification, validation and authentication. Developers usually implement some sort of data “scrubbing” routines into the application server’s processing to eliminate the presentation of duplicate records, incomplete records or NULL results.

Application Server Software

There are two major contenders in the application server software market: Java (Oracle) and .NET (Microsoft). The Java application server is a cross-platform language and runtime environment, which means that it is platform independent and that it maintains compatibility with Windows, Linux, UNIX, Mac and other server platforms. Microsoft’s .NET only operates on the Windows operating system although there is a project currently underway whose purpose it is to port .NET applications to Linux.

The Advantages of Well-Designed Architecture

Although anyone can find numerous examples of Three-Tier designs and Web Application How-Tos on Internet sites, there’s no substitute for a professionally crafted data-backed web application. To maintain a web application infrastructure, requires a trained team of IT professionals including: System administrators, DBAs and application developers. But, no matter how good your support staff is, a poorly architected web application solution will never provide you with the service you expect from it. In a Three-Tier web application, make sure that you have an adequate number of web servers available to accommodate the amount of traffic you expect because your web servers will be very busy.

Some architects use a combination of physical web servers and virtual web servers that administrators spin up to adjust for high traffic times (during special promotions, for example).

Apply weighted load balancing to your web servers and to your application servers. Also enable session affinity (sticky sessions) in your load balancing setup. Using session affinity at this level greatly simplifies some of the session management in the application.

The Pre-Production To Do List

During the pre-production phase of your web application launch, a few things need to happen. The first is load testing. You need an experienced load tester to place stress on your system to make sure that it can handle many simultaneous users. On the user interface side, you should enlist a software tester to ensure that your interface is intuitive and not easily broken by erroneous input. Additionally, you need a representative group of users to provide feedback on the user interface. Finally, you should have a security audit performed on the environment to include penetration testing and vulnerability testing.

Summary

When the need arises for your application to go public or to reach a large audience, you need to move to a scalable and manageable architecture. Three-tier architecture is one very good answer to that problem. 3TA is true data center architecture that includes a security component, a performance component and an availability component. Put them together and you’ve built a near-unbreakable service for your intended user base. The best web application service begins with exceptional design and ends with happy customers.

This post was written as part of the IBM for Midsize Business program, which provides midsize businesses with the tools, expertise and solutions they need to become engines of a smarter planet.