Integrating Cassatt Active Response into Your Enterprise: Automation at Your Own Pace
Intended for use with Cassatt Active Response V5.0
Cassatt Active Response offers a broad range of control over your computing
environment to ensure that your business applications run
at guaranteed, user-defined service levels. Depending on
the level of Cassatt Active Response control you are comfortable with, the
service-level agreements (SLAs) you need to maintain, and
the ROI you want to achieve, you can choose the level of
integration appropriate for your site. You can integrate
Cassatt Active Response seamlessly into your environment, you can modify
your environment to maximize use of your server resources,
or you can mix simple and complex implementations, depending
on your applications.
In this article, I'm going to outline the range of Cassatt Active Response implementations
and highlight both practical and technical factors to consider
when choosing one. Let's start by considering various modes
of operation.
Modes of operation
In its simplest
sense, the range of Cassatt Active Response control looks something like
this:
I'm using
the term mode here
in a general way. Cassatt Active Response doesn't really have a toggle switch
you can flip from one mode to another. Rather, these are
end points on a continuum. (In
fact, you can run in multiple modes at one time.) For the sake
of discussion, though, let's look at various Cassatt Active Response features
as they relate to these end points.
| |
Least automation |
Most automation |
| Maintains application service-level
agreements |
 |
|
| Can reboot nodes on failure |
|
|
| Can bring new nodes online
to assume workload of a failed node |
* |
|
| Can notify users about server
events such as a failure |
|
|
| Provides utilization reports |
|
|
| Powers down servers not
in use |
** |
** |
| Supports any operating system |
|
|
| Can automatically discover
and inventory servers as they are introduced into the
environment |
|
|
| Dynamically pairs applications
and computing resources |
|
|
| Dynamically configures network
switches so that servers can be allocated to networks
where computing resources are needed |
|
|
| Manages VMs |
|
|
| Maintains application versioning
in a common image repository (the Cassatt Active Response image matrix) |
|
|
| Manages update of applications
without interruption to application operations |
|
|
* Limited to fixed set of resources.
** Available on a range of supported power management devices.
As I mentioned, it is not an either/or proposition to implement Cassatt Active Response
for basic SLA management or for utility computing.
You can integrate Cassatt Active Response in
a manner consistent with your business goals and tolerance
for change. Let me describe how.
top
Integrating Cassatt Active Response seamlessly into your environment
If your goal is to use Cassatt Active Response to maintain application service
levels for your user community with the least amount of
change to your existing environment, you can integrate Cassatt Active Response
seamlessly—in most cases you
don't need to modify your existing servers or business applications
at all. You simply install and configure Cassatt Active Response with the
IP addresses of the servers to control and
specify the service-level requirements, defining the number
of servers optimally and minimally required to maintain
service levels.
Manually allocated resources
Cassatt Active Response uses an organizing mechanism called a tier to
define the requirements for managing each application.
Tiers are configured with different parameters
for different purposes: the main parameter of interest
for this discussion is whether the tier's application
servers are manually specified and dedicated to
an application or dynamically allocated and shared
with other applications.
When servers (specified by their IP
addresses) and their accompanying applications are
statically assigned to Cassatt Active Response, they are called manually
allocated resources. Cassatt Active Response simply
maintains SLAs on the set of servers you manually
specify and the applications already running on
those servers. In Cassatt Active Response, this
is generally referred to as a manually allocated
tier. (For the detailed setup steps,
see Manually
Allocated Tiers: Configuration and Management.) |
In a manually allocated tier, Cassatt Active Response uses
Secure Shell (SSH) and access to the remote management controllers
on the servers it is managing to gracefully bring them up or
down. Cassatt Active Response
maintains the service levels you've specified,
ensuring operational target behaviors are
maintained. If
a server fails, Cassatt Active Response automatically powers up and boots
another node into service. This is baseline Cassatt Active Response functionality,
and it requires no change to your operating environment.
Furthermore, besides guaranteed service levels, it provides:
- Ability to automatically reboot servers on failure
- Automatic event notification
- Utilization reports
Cassatt Active Response requires that the application
server hardware use supported remote management controllers
so that Cassatt Active Response can access
power operations on those servers. Supported onboard remote
management controllers include Dell
Remote Access Controllers (DRAC), HP Integrated Lights-Out
(iLO), and IBM Remote Supervisor Adapter II (RSA II) or
BladeCenter Management Module (BCMM). Supported external
power distribution units include those from APC, Baytech,
and Dualcom.
top
Moving to utility computing
If your goals are not only to guarantee service levels, but also
to increase server utilization, decrease administration costs,
and decrease capital expenditures, then you are a candidate
to move to a Cassatt Active Response utility computing model. However, running
your IT as a utility requires not only that Cassatt Active Response know
about your computing environment, but also that your computing
environment be prepared for Cassatt Active Response control, so this is
a more pervasive implementation of Cassatt Active Response. Generally speaking,
to support Cassatt Active Response utility computing, your server and application
environment needs to be modified to allow Cassatt Active Response to discover
and inventory servers, boot them up/shut them down based
on demand, and deploy application images. You
may also want to allow Cassatt Active Response to manage your network switches
for the purpose of allocating servers to networks where computing
resources are needed. Specifically, a utility
implementation of Cassatt Active Response requires the following enabling
technologies and practices:
- Allowing Cassatt Active Response to be the DHCP server for
the environment – In
a utility computing implementation, Cassatt Active Response acts as the
DHCP server. Cassatt Active Response answers DHCP requests
and uses this information to discover and pool server
resources together for use in the Cassatt Active Response environment.
- Enabling PXE booting on application servers – Cassatt Active Response
uses Preboot eXecution Environment (PXE) technology to
boot application servers via their network interface cards
(NICs). Setting this up usually requires configuring the
application server's BIOS and remote management controller.
- Capturing application images and storing
them in the Cassatt Active Response image matrix – Cassatt Active Response leverages its image matrix to serve
applications dynamically to nodes available in the free
resource pool.
- Allowing Cassatt Active Response to dynamically
configure network switches – Cassatt Active Response
can use its Network Virtualization Service (NVS)
to shift servers into and out of networks based on
application requirements. For example, some applications
may require isolated, firewalled networks,
while others may require secondary, dedicated networks
for application-specific communication.
These technologies and practices are key to enabling dynamic and automated
control of the environment so that Cassatt Active Response can transparently
deliver appropriate resources wherever and whenever
they are needed.
Dynamically allocated resources
Again, in Cassatt Active Response, you organize applications and
define their SLAs in conceptual groups called tiers.
In contrast with the simple implementation I described
earlier, in which Cassatt Active Response operates on a manually
specified set of servers and their accompanying
applications, a utility implementation of Cassatt Active Response
provides automated and dynamic control over the
environment. This configuration would be considered
a dynamically allocated tier in Cassatt Active Response.
The procedures for such a configuration span multiple
documents. For an overview of tasks, see the Implementation Task Flow: Premium Edition, Data Center Edition. |
top
Using DHCP to inventory and pool computing resources
You can either manually specify which servers you want Cassatt Active Response to control,
or you can use the Cassatt Active Response auto-discovery feature. With
auto-discovery enabled, Cassatt Active Response automatically registers
the availability of servers and places them in the Cassatt Active Response free resource
pool.
It works like this. Cassatt Active Response is configured by default as a DHCP server.
When Cassatt Active Response auto-discovery is enabled, Cassatt Active Response responds
to DHCP requests from any supported remote management device
(see Recommended Hardware),
assigns an IP address, and inventories the server hardware.
This inventory information can then be used for cases where there are
minimum server hardware requirements to run certain applications; for
example, say an application requires a minimum amount of RAM. Cassatt Active Response
can then match servers to applications based on the applications'
hardware memory requirements. Once
servers are inventoried and placed in the free resource
pool, Cassatt Active Response dynamically pairs them to applications
that require computing resources within the Cassatt Active Response environment.
Allowing Cassatt Active Response to answer DHCP requests is required for auto-discovery
of server resources and is a feature of a fully dynamic
environment. This is one of the ways Cassatt Active Response repurposes
existing server/application silos for the sake of utility
computing. However, if you do not want to use the Cassatt Active Response
DHCP services, you can disable auto-discovery and manually
specify the servers to be under Cassatt Active Response control. (By the
way, the manually allocated tier implementation
I mentioned earlier requires that auto-discovery is disabled.)
For more information on how Cassatt Active Response auto-discovery works, see Understanding
Automation: Node Discovery, Inventory, and Quarantine.
top
Using PXE to boot application servers
Enabling PXE on application servers gives Cassatt Active Response the means to
boot them over the network. In the case of a manually
allocated tier discussed previously, PXE is unnecessary
because Cassatt Active Response just issues commands to the
server's remote management controller to power it on/off
and issues commands via SSH directly to the server to
achieve an orderly shutdown. However, in a utility computing
environment, where Cassatt Active Response manages multiple applications
contending for resources from a common resource pool, the
PXE technology is used to bootstrap servers with an application
image from the Cassatt Active Response image matrix (which I'll talk about
next).
top
Serving applications from the image matrix
As I described earlier, in a Cassatt Active Response manually allocated tier implementation,
Cassatt Active Response has no direct knowledge of the applications running
on the servers under Cassatt Active Response control. In a utility, dynamically
allocated tier implementation, however, Cassatt Active Response
hosts those applications in its image matrix. Cassatt Active Response can
then allocate servers to applications at runtime based on
user-defined service-level agreements. Cassatt Active Response serves these
applications in one of two ways: NFS mounting the application
image on the server or by copying the application image
to the server's disk.
top
NFS mounting application images from the image matrix
Mounting application images via NFS
is the preferred use of the image matrix. From the image
matrix, Cassatt Active Response can manage and ensure consistency of application
updates, apply updates to application tiers without disrupting
operations, and maintain application state in the case of
server node failure. Using NFS provides the most automated
and robust solution for Cassatt Active Response utility computing.
top
Copying application images from the image matrix
Sites averse to the possibility of network connectivity problems affecting
application uptime can still leverage Cassatt Active Response's dynamic
allocation of computing resources. Rather than serve the
application images via NFS, Cassatt Active Response can
copy the application image from the image matrix to the
server's local disk. However, this method
does come with a downside. Runtime data that
is written to disk from these applications could be lost
in the event the server goes down. For example, if a server
crashes and Cassatt Active Response reboots it, but that doesn't correct
the problem, Cassatt Active Response will then bring up a new server and
copy the application from the image matrix to the new server.
In this case, data on the server that went down is lost. (Alternatively,
when NFS mounted, the application runtime data is written
back to the image matrix, so Cassatt Active Response preserves the exact
state of the application if the server goes down.)
top
Using network virtualization
In a utility computing environment, you may want
to leverage your application server pool independent of
an application's network requirements. If so,
you can use the Cassatt Active Response Network Manager, which decouples your network configuration from
your application services. Cassatt Active Response achieves this by exercising
programmatic control over server network connectivity.
The result is that Cassatt Active Response can allocate servers automatically
to applications that have specific networking requirements—without
regard to the servers' current network connections. Cassatt Active Response
simply selects a server with sufficient NICs from the
global resource pool and reprograms its switch port connections
to meet the application's networking needs.
top
Mixed-mode operation
I've described Cassatt Active Response mostly at its extremes—as a basic implementation
that integrates seamlessly into your environment or as
a utility computing implementation that requires some modification
to your existing environment to facilitate automation and
dynamic resource allocation. However, all of these flavors
of Cassatt Active Response are available in one package; you can configure
both types of implementations within one Cassatt Active Response domain.
Doing so makes sense if you:
- Have different automation goals
for different applications
- Have different demands
for containing costs for different departments
- Want to use basic Cassatt Active Response SLA capabilities on applications
that run on operating systems not yet supported in a utility
implementation
- Want to try Cassatt Active Response in a simple implementation and migrate over
time to a full utility computing implementation
For a more exhaustive comparison of features, functions,
and configuration requirements for running applications
under Cassatt Active Response control in manually allocated tiers or dynamically
allocated tiers, check
out this
table.
top
Extending Cassatt Active Response via scripting
You can extend Cassatt Active Response and achieve even greater levels of automation
by using the Cassatt Active Response Scripting Software Developer's Kit
(SSDK). The SSDK provides a scripting interface to various
Cassatt Active Response operations.
Even if you are starting with a simple Cassatt Active Response implementation,
you can bring a sophisticated and customized degree
of automation to your environment by leveraging the SSDK.
For example, say you wanted to power down servers on the
weekend to save on electrical costs, or you wanted to reallocate
servers based on scheduled demands on computing
capacity. For some examples, see Scripting Scheduled Configuration Changes for Cassatt Active Response and Cassatt Active Response Scripting SDK Command Reference.
top
Related documentation
There are several other documents that describe procedures for the
various implementations I've been talking about. Here are
a couple to take a get you started:
top
Conclusion
Cassatt Active Response implementations span very simple, seamless integrations with
your existing environment to more pervasive control of that
environment. However, you can choose an implementation that
makes sense for your goals, desired ROI, and tolerance
for change. Baseline Cassatt Active Response functionality guarantees services
levels with no change to your existing environment. Increasing
server utilization and decreasing administrative and capital
costs requires that your environment be prepped for Cassatt Active Response
automation and dynamic allocation of resources to meet demand.
The more control Cassatt Active Response has of your computing resources
and application images, the greater the utility computing
benefits.
top
Was this article useful? Tell us what you think.
Email infocentral@cassatt.com.
|
