Network Addresses: Calculating Requirements

Intended for use with Cassatt Active Response V5.0.

When you install the Cassatt Active Response software, the installation program configures the control node to act as a DHCP server. In this way Cassatt Active Response manages IP allocation to items under its control. So, when planning for a Cassatt Active Response implementation, you need to determine the IP address pool for servers, network devices, and applications participating in the Cassatt Active Response environment. To this end, I'm going to show you how to:

• Calculate IPs required for the main network under Cassatt Active Response control, the Cassatt Active Response network.
• Calculate IPs required for additional networks that may be required by your site networking policy.
• Reserve IPs that need to be reserved (because some devices require static IP addresses).
• Determine a range of contiguous IPs in the Cassatt Active Response network that Cassatt Active Response can use to dynamically assign to items under its control. (You need to have this IP range in hand before you install Cassatt Active Response.)

top

What you should know

In some ways, determining an IP range for a network in Cassatt Active Response is similar to any network planning exercise. For example, traditionally, you would determine your required IP address pool by adding the number of servers you expect to reside on the network, the number of miscellaneous devices, and then pad as you see fit. Voila, you have your IP range and, after a bit of binary math, a netmask for your network. However, Cassatt Active Response adds a few variables to the mix. To calculate IP address ranges for Cassatt Active Response, I'm going to walk you through the following two-step formula:

1. Calculate IPs required by the Cassatt Active Response network, which contains the shared resource pool
2. Calculate IPs for additional networks

Before I go into the details of each calculation, I need to mention briefly the network strategies available to you and their relative pros and cons.

top

Network strategies

Cassatt Active Response allows you quite a bit of flexibility in designing your networks. You can determine the IP address pool based on either one large Cassatt Active Response network, or determine multiple IP pools for multiple networks. The following table outlines the relative advantages/disadvantages of each approach:

For more information on the range of networking configurations supported in Cassatt Active Response, see Understanding Cassatt Active Response Network Manager.

top

IPs for the Cassatt Active Response network

At a minimum, Cassatt Active Response requires a network that includes a pool of servers that can be allocated to applications running in the Cassatt Active Response environment. The Cassatt Active Response network must have enough IP addresses to account for:

• The Cassatt Active Response control node(s)—you can have one or two.
• Any other device connected to the network, including a gateway, which is required for the Cassatt Active Response network.
• Each physical server that will participate in the Cassatt Active Response environment.

The only thing that's not immediately obvious is that, when using two control nodes, you need to factor in an extra IP (actually, a virtual IP) that is used to support one control node failing over to the other. That's not so unusual, is it?

Now let's look at some of the IP address requirements for additional networks you might want Cassatt Active Response to supervise.

top

IPs for additional networks

It's common in IT to dedicate networks to applications or other functions. In Cassatt Active Response terms, this means creating networks (and determining the necessary IP pools for them) for application tiers. An application tier is a logical construct in the Cassatt Active Response environment defined to represent an application and its physical and virtual resources. For example, you might have a database tier, a web server tier, or any other application-specific tier. The idea is that, based on how you define the tier, Cassatt Active Response allocates computing resources (and IPs, as necessary) to meet your service level requirements. You give each application tier a max nodes value in Cassatt Active Response. Cassatt Active Response then reserves an equal number of application image instances, each of which gets an IP address. (I'll explain more about image instances later, as it is important to your calculation.)

In terms of determining the IP addresses necessary for these additional networks, the calculation is a little different from the calculation for the main Cassatt Active Response network. In general, these networks require an IP for each virtual server, application image instance, and miscellaneous device on the network. (Actually, it's a little more complicated than that. The nature of running some applications mandates that extra IPs are available, but I don't want to bog you down with those details yet.)

top

Determining IP address pools

Okay, now that you are either 1) suitably enlightened about networks and IP addressing in the Cassatt Active Response environment or 2) bored with the background info and ready to get down to business, I'll walk you through the formulas to determine your IP address pool or pools.

Step 1: Determine IP addresses and netmask for the Cassatt Active Response network

First, calculate the number of IPs you need for the Cassatt Active Response network. This is relatively straightforward. In general, though, consider these guidelines:

Now, specifically, to calculate the number of IPs required, use this formula:

Where:
(number physical servers)
This is the total number of physical servers you expect to run in the Cassatt Active Response environment. Cassatt Active Response allocates IP addresses to them for the purposes of inventorying them and placing them in the free pool. (Include the number of blade servers in a blade enclosure.) You'll want to pad this number for anticipated growth. Also include the number of control nodes in this total. You can have either 1 or 2 control nodes.

(number of other IPs required on the network)
Include gateway IPs, firewalls, load balancers, sniffers, and virtual IPs (VIPs). Remember that if you use two control nodes, you must have a VIP. Also, if you use redundant gateways, add a VIP. The Cassatt Active Response network requires a gateway, so the minimum value for this variable is 1.

(padding)
Padding should minimally be set to 10 for the Cassatt Active Response network.

This formula gives you a total number of IP addresses needed for the Cassatt Active Response network.

top

Example—calculating the Cassatt Active Response network IP pool

Assume a simple configuration that includes the following:

• 26 physical servers (including 2 control nodes)
• 2 routers serving as redundant gateways

Given this configuration, I'd calculate a minimum IP pool like this:

Of course, pad the total as you see fit to anticipate an increase in physical servers.

That's it for the Cassatt Active Response network. Note that you cannot easily increase the IP range for the Cassatt Active Response network after Cassatt Active Response is up and running, so think through this calculation thoroughly. If you intend to have one large network, complete the second part of of this calculation to make sure your IP range accounts for power controllers on the network.

top

Step 2: Determine IP addresses for other networks

Next, determine IP addresses required by additional networks. (Remember, I said that the same calculations will work for single or multiple networks, so if you prefer one king-sized Cassatt Active Response network, follow along. You need to do step 2 to determine additional IPs for the Cassatt Active Response network. I'm just biasing the discussion for multiple networks.)

With the advent of virtualization—a server may need an IP for each VM—calculating network size becomes more complex. But more and more it's the calculation that IT network architects have to make. Similarly, in Cassatt Active Response, you also need to account for application image instances you'll be running in the Cassatt Active Response environment. Both virtual servers and application image instances require IP addresses.

Special network multipathing (as in IPMP), clustering software, or VMs on a primary network also require extra IPs. I'll point out the details below, but let's start with the general formula:

Where:

(number power controllers)
This is the number of power controllers on the physical servers you expect to run in this network. Cassatt Active Response assigns IP addresses to these and uses them to power cycle servers. Many sites and some hardware vendors recommend that these controllers be on their own network for security reasons. If you want a network dedicated to power controllers, just total the number of power controllers and you're done with the calculation for this network. In a dual-control node configuration, be sure to include those power controllers in the total.

I'm going to refer generically to a server's remote management controller as a power controller because Cassatt Active Response accesses remote management utilities primarily to perform power on/off operations.

(number application image instances)
Cassatt Active Response dynamically provisions bare-metal servers with application image instances from its image matrix. Each image instance requires an IP address. This value is the maximum number of application instances you expect to run on this network. (This equates to the maximum number of servers required to support each application that will run on the network.) For example, if you expect the maximum number of Apache application instances to be 8, then enter 8. However, to the total number of application instances, apply these factors, as necessary:

Your application may have its own special IP requirements, so consider those as necessary.

(number of virtual machines [VMs])
This is the total number of VMs you expect to run on this network. However, only add this number to your total if this network is the primary network for the virtual machine manager (VMM) application tier (I'll show you what I mean in the ESX Server example below). If this is not the primary network for the VMM base tier, skip this input into the calculation. (You need to know whether each network is the primary network for a VMM base tier because this is the network that provides IPs to the VMs generated by the application.)

(number of other IPs required on the network)
This is the number of other devices that require an IP address: for example, gateway IPs, routers, firewalls, load balancers, sniffers, virtual IPs, etc.

(padding)
As before, padding should minimally be set to 10.

If you intend to add more application networks, repeat step 2 for each network. And if you want to configure just one network (which would host the shared resource pool and all application tiers), just add the totals from step 2 to the total from step 1. That gives you the minimum number of IPs for one large network. In the end, you're probably going to need a serious spreadsheet to plan, define, and calculate IPs for these networks. I've included one you can use at the end of this article.

top

Example—calculating additional network IP pools for VMware ESX Server

One of Cassatt Active Response's features is its ability to manage both physical and virtual servers in a platform- and vendor-neutral way. Because I need to consider VMs in the IP calculation, let's look at an example that accounts for a major server consolidation effort utilizing VMware ESX Server. (See the ESX Server 3.0 blueprint for instructions on preparing the VMware ESX Server 3.0 software to run in the Cassatt Active Response environment.)

For the base tier primary network, I need to account for:

• The the maximum number of ESX Server application image instances
• A second IP address for each of the ESX Server application image instances (this IP address is required by ESX Server for its own use)
• The number of VMs configured in each instance
• A gateway/firewall

Here is how I visualize the network:

Notice when you get to the calculation table below that this illustration shows 12 virtual servers (3 application image instances, each configured with 4 VMs). I'll need IP addresses for all 12.

Because this network will also be assigned to the IIS guest application, I also need to add the IIS application image instances that will run in the VMs. For this example, let's assume I want to run IIS in the VMs. Let's suppose I want the maximum number of IIS image instances to be 6. I'll need to add 6 more IPs for this network:

So, this is how my IP calculation unfolds:

top

Example—calculating additional network IP pools for Oracle 10g RAC

For the sake of this example, assume we're configuring Cassatt Active Response to run Oracle 10g RAC. Now Oracle 10g is a complex application, and I know from reading the Oracle 10G blueprint that in a Cassatt Active Response implementation it requires 2 networks:

• A primary network from which Cassatt Active Response serves application image instances
• A network for the Oracle 10g application cluster

My configuration is going to look something like this:

So I need to calculate IPs for both networks. Let's start with the primary network.

Oracle 10g primary network

Cassatt Active Response uses the primary network to boot hardware, serve application images, and monitor the application tier. For this calculation, I'll factor in the following:

• 3 application image instances
• VIP addresses, one per image instance, as required by Oracle 10g RAC
• 1 gateway/firewall

Here's how my IP calculation looks:

Now let's look at the requirements for the Oracle 10g cluster network.

Oracle 10g cluster network

The dedicated network, where a physical NIC is allocated for use only with that network, allows for communications within the Oracle 10g application cluster. It's going to include:

• 3 application image instances
• 1 gateway/firewall

The cluster network IP calculation looks like this:

So, for Oracle 10g, I have two networks, each with its own IP address range:

Reserving static IPs and determining IP dynamic range for the Cassatt Active Response network

After you've calculated the IP address pool for the main Cassatt Active Response network, you need to divide it into:

• One or more sets of IP addresses that are reserved for devices that require static IPs
• A range of contiguous IP addresses that Cassatt Active Response dynamically allocates to devices

You must reserve static IPs for the Cassatt Active Response network prior to installing Cassatt Active Response. The Cassatt Active Response installation program prompts for this information. Cassatt Active Response then uses it to determine which IPs it can dynamically assign. If Cassatt Active Response were to dynamically allocate an IP address that you have otherwise statically assigned, you would end up with undesirable system behavior. I'd recommend you separate these now as part of your network architecture to avoid that possible problem.

top

Reserving static IPs

Cassatt Active Response requires that a few devices have static IPs:

• Control nodes and their VIP
• Gateways and their VIP
• BladeCenter integrated switch modules

Typically, it's common networking practice to reserve static IPs from the start or the end of the IP range:

top

Determining the dynamic IP range

After you have identified the reserved IPs you need from the start and end of your total network space, you are left with a range of contiguous IP addresses. These are the ones Cassatt Active Response dynamically (the D in DHCP) allocates. One important thing to remember about this range of contiguous IPs: during the Cassatt Active Response installation, you need to specify the first available IP and the last available IP in that range.

At installation time, the first and last available IP values indicate the range of IPs available for dynamic allocation. Note that after Cassatt Active Response is up and running, you can extend this range by reserving IPs (from anywhere in the IP range) in the Cassatt Active Response Controller user interface. And when you create additional networks, you also use the Controller to reserve IPs.

Separating the IP range into static and dynamic groups isn't conceptually that difficult, but let's look at an example to make it more concrete. This time, I'm going to complicate things by significantly increasing the size of our example network.

top

Example—determining static IPs and range of dynamic IPs

Let's assume I am calculating an IP range for my Cassatt Active Response network. In this example, I'm going to put a large server farm under Cassatt Active Response control. (I intend to add separate networks for various applications, so this exercise focuses on the Cassatt Active Response network. However, I'll include the power controllers on the Cassatt Active Response network for this example.) Here's how I see my network taking shape:

For padding, I'm rounding up to a 1024 boundary, but for practicality, I've subtracted 2 from that range to account for 1 broadcast and 1 network address, which cannot be assigned to devices. Assume that, as network administrator, I determine the network address is 10.10.80/22.

Now, Cassatt Active Response requires that I set aside a few IPs for devices that need static IP addresses. To that list, I'm going to reserve a few extra for miscellaneous devices and future use, so I'm going to reserve 12 IPs from the beginning of the IP range and 12 from the end of the IP range. My list of reserved IPs looks like this:

After I separate the reserved IPs from the beginning and end of the total IP range, I'm left with a range of 10.10.80.13–10.10.83.241 that Cassatt Active Response can use to dynamically assign to resources under its control. These, then, are my first available and last available IPs, which I'll need to provide when I install Cassatt Active Response on the control nodes.

top

Netmask lookup table

You are going to need a netmask for each network. If you really love your binary math, you can determine a network mask for each network based on its respective pool of IPs. You'll need the netmask for the Cassatt Active Response network when you install Cassatt Active Response, and for additional networks when you configure application tiers in the Cassatt Active Response Controller user interface. However, I'll provide a little cheat sheet for you.

Once you've calculated the number of IPs you need and determined a netmask for each network, you can get your network addresses from your admin.

top

Network address planning sheet

We've largely worked through a big planning exercise in this article. After you've done the work of determining various network IP pools, netmasks, static IP assignments, et al., it's a good idea to capture that information. Click this link for a handy little planning sheet that can be used during the various steps of a Cassatt Active Response implementation:

Network address planning sheet

As your site personnel go through the various Cassatt Active Response setup and configuration phases, they'll thank you for completing this planning sheet (and you'll probably be happier with the results, too).

top

Summary

Remember that determining one or more IP address pools requires you to:

1. Calculate IPs required by the Cassatt Active Response network, which contains the Cassatt Active Response shared resource pool. Divide the Cassatt Active Response network into a contiguous range of IPs for dynamic allocation and a set of reserved IPs for static allocation.
2. Calculate IPs for additional networks

Even if you only want one large network, you can follow these two steps and then add the results together to calculate a minimum IP address pool.

I've covered a lot of ground in this article, so I'll summarize some of the key points about network addressing in Cassatt Active Response in this table.

top

Was this article useful? Tell us what you think.
Email infocentral@cassatt.com.