Birds-Eye.Net DHCP Written Works

Check out these other Birds-Eye.Net papers/products regarding DHCP:

Products:

• DHCP Stress Testing Suite - The only commercially available DHCP stress testing suite
• DHCP Snoop Agent - The perfect tool to count various types of DHCP packets and more...
• DHCP Custom Packet Generator - A one shot tool for composing, sending, and receiving individual DHCP packets

White Papers and Reading Material

• ABCs of Understanding DHCP Performance - A beginners guide to understanding operation and performance of DHCP servers
• Carrier Class DHCP Testing - Devising a carrier-class test environment to measure near-real-world DHCP performance
• DHCP Filtering - Stopping unwanted DHCP packets before they can damage performance
• DHCP Network Traffic Analysis - What represents "normal" traffic on a large DHCP network?
• DHCP Recommendations - Insight into the first major DHCP deployment and what was going through our heads
• IP Address Management - An introduction to Internet Protocol (IP) address management
• IP Address Provisioning Overview - An introduction to IP address provisioning for data over cable service

Meta Research:

The enterprise DHCP market represents an ever-growing opportunity for high performance software applications that can address needs of rapidly expanding businesses with numerous remote offices and global operations. All businesses today utilize DHCP as a standard networking application to facilitate IP address distribution and management. However the DHCP application needs for these companies vary greatly depending on the relative size of the company.

Dataquest forecasts that the worldwide server appliance market will be $16 billion worldwide by 2003. With all the different types of servers, DHCP accounts for between 1/8^th and 1/10^th of this appliance market (a rough but conservative estimate) or about $1.6-2 billion. Dataquest does not break out enterprise vs. Internet service or software vs. hardware so it is believed this represents the total market.

Enterprise DHCP servers typically sell in blocks of 1000 nodes. With initial server costs (plus licensing for the 1000 nodes) start at around $5-7k for advanced servers, to around $9-10k for entry level high availability systems, pricing above $10k is unavailable (publicly) but probably does exist for premium server software but the market for that type of software is considerably smaller than that of your average fortune 500 company.

For $10k (software only) your going to be able to get a fairly sophisticated server complex that supports up to 1000 nodes, a number of enterprise specific analysis and reporting tools, DDNS update capabilities, Microsoft Active Directory Server interoperability, centralized control, and distributed redundant servers (that support DHCP fail over). Pricing per node after the initial $10k drops off drastically approaching and going under $1 per node in some cases for bulk purchases. These prices are generally not listed and ‘highly’ negotiable.

Desired features of enterprise DHCP servers

Features:

• delivery of information such as news, dns, time, printer, WINS,
• Out-of-band 2way communication channel for changes to configurations - configuration on-the-fly
• configure all operating parameters over the network
• add, modify, delete lease data over the network
• Must support frequent changes to network structure -- both local and remote
• User-defined client attributes for policy implementation
• Database configuration for lease pools and network/configuration data
• DHCP-INFORM support

Performance & Scalability:

• Fast start and restart - restarts to a few seconds
• Optimized and compiled configuration database structure that does not need to be loaded at restart?
• process of handling new leases is separate from that of handling synchronization of existing leases, Separate processes mean shorter wait times between to obtain both new and renewal leases
• Low memory consumption
• Failover support such that yields ZERO client downtime and can leverage dual server "load balancing"
• Scalable to hundreds of thousands of nodes
• Support for Safe DHCP failover

Security:

• Security of the database - information is protected

Administration and Access Control:

• Delegating administrative control to personnel with DHCP servers at remote locations
• Can be run with user privileges rather than root (or admin)
• Log all events that modify the configuration or lease databases
• Role based and rules based configuration
• Securely work with Windows Active Directory services
• Analysis tools for troubleshooting and reporting - for example some tools can locate rouge DHCP servers automatically and that all DNS configurations and zones are error-free. Manual or scheduled run for reporting/analysis.

Analysis Tools:

• Locate rouge DHCP servers automatically
• Ensure that all DNS configurations and zones are error free
• Manual or scheduled runs
• History of IP address assignments and configuration changes made by remote administrators
• Asset tag tracking

Enterprise DHCP server vendors:
 

Factors that impact how many DHCP servers to use:

WAN links and connectivity -- distant locations, those with no redundant connectivity, or those with very slow redundant links should maintain their own DHCP server.
Disk drive speed -- faster the better
number of scopes -- best if kept under 1,000 total
Keep total clients per server under 10,000 (Microsoft Recommendation)

Links to Enterprise DHCP server articles, white papers, etc.

Nominum White paper on Enterprise DHCP - interesting albeit slightly biased  

Notes from discussions with DHCP vendors:

Enterprise vs. broadband 

QIP: 40 % of marketshare, enterprise space, wireless, and wireline Tim Rooney, common denominator is DHCP also DNS. Why can’t I use a basic DNS and DHCP server? From an enterprise level – need to be able to manage IP addresses for multiple regions. Maintain precious resources (configurations, IP addresses). Windows enterprise 2000 functionality support. Vital access provides a hook into QIP to provide broadband functionality – however the basic requirements of enterprise and broadband DHCP requirements are identical. 

Netid: primary focused towards the enterprise market. Some features (like security, etc.) of the broadband space are unique to broadband. Different types of features specifically address broadband features. Broadband (ISPs) generally have their own backend systems and want to tie their own backend into their DHCP markets. Never had a request for employee provisioning as part of enterprise requirements. NetID does some employee provisioning – through client side provisioning (bubbled up through employees setting hostnames on their OS client configurations). Integration through windows 2000 servers – value proposition or enterprise requirements. When you bring in windows 2000, you have this service registration requirements. Each configured service will want to register this service name with the DNS server. Enterprise (anti-requirement) for APIs. They want GUIs for system control rather than APIs (such as in the case of broadband). 

Most enterprises are using Microsoft DHCP. Does come with NT or 2000 server. Company that would buy a server would have a number of locations to support before they could make this pay off. Still need to have the tools and the access.  

#1 Microsoft does not have fail over
#2 does not have web management (remote capabilities)
#3 diagnostics are very weak
#4 limited auditing information – only see active addresses 

If your going to manage address space from a remote location. Granular access. Can assign administrators to particular address space (scopes). ACLs allow scope based access control. Regional list. Both view and control. 

Support for DHCP is sorely lacking with respect to Microsoft DHCP. 

Value proposition? 

QIP: Interaction with customer is unique. Labor savings. Automation. Using QIP or not using QIP. Minimum 2 administrators, to 100 administrators overseeing 1 million addresses. Performance, scalability, flexible manage can interface with other non-lucent products. Administration is very powerful and more feature rich than competition. Windows 2000 support goes beyond competition. Couple customers running 5 million addresses, couple universities running with a 1000 addresses. Largest single server supporting 300-400 thousand IP addresses. 

NetID: There is a need to provide a centralized IP address management system. Use BIND (open source DNS server). Enterprise does see BIND but don’t see how it gets configured. Want to manage domains as domains rather than configuring these services individually. Don’t have to worry about hand editing configuration files for DNS or the errors associated with it. Can delegate administration. Win the most deals with – GUI is very strong. View configuration as a single system, so have consistency across multiple. Workflows seem to be simpler. Security features are what they are looking for. Do not present an ROI. View on the ROI, very tough to come up with a generic one. So many ways that IT houses are structured. Try to focus on what the value proposition is and let the customer decide which items are more important. 

IN: ROI calculator for IP management. Minimize administration. Fewer administrators and fewer time spent managing product, also less time spent diagnosing problems. Some companies require an audit trail of IP address use. 

How large do you perceive your market being? 

QIP: no comment 

NetID: Primarily fortune 1000 companies. Get into specific segments outside of fortune 1000 such as government, university, healthcare, and banks. Parallel segments to fortune 1000 as well as synonymous. Sized it up a number of different ways. From100 million to 500 million. A lot of cross over for smaller ISPs. Carrier and large ISP side to provide IP address service as a “service” rather than selling software. 

IN: Longer sales process than the broadband business. It’s a nice to have as opposed to a must have. Tom Larkin would have numbers on market size. Universities are a BIG user of self provisioning system – one sector that DHCP is a necessary environment also universities are also not very Microsoft focused so that also generates into heightened interest in must have DHCP products. 

Who do you view your competition to be? 

QIP: Cisco (CNR mainly on the service provider side), checkpoint (lower end enterprise), NetID (fell by the way side – hasn’t seen a new release 18 months ago – V4.2 – base DHCP on ISC). Incognito – more actually on the service provider space – more likely on the service provider space. 

NetID: 3 or for players. CheckPoint, Cisco, Lucent. A number of smaller players. Smaller players provide a naming IP address market. There are a number of fortune 1000 companies already have the management, so all they need is the server software rather than a centralize product (out of the box). The other areas they are seeing is 1U box that addresses the enterprise requirements, but still need remote resources to configure. Lucent, NetID, Cisco, then CheckPoint. 

IN: Lucent vital access (has the largest share), Infoblox (hardware only product), Nominum (more in the DNS space, have yet to see a DHCP product come out), Check Point MetaIP, occasionally Cisco. 

What are the factors impacting your potential customers from purchasing your product? 

QIP: Pricing/Budget. Customers can get ISC for free. Cisco can provide CNR for discount with purchase of routers. Need administrator flexibility. Custom development – but is rare. One add on is an API toolkit – some customers want to modify that. 

NetID: Pricing is the biggest obstacle. Doesn’t tend to be a competition thing. Sales may have more information on sales process and the factor that budgets plays. 

IN: Budget, where you fall in the budget, really showing the difference between you and Microsoft are worth the money. Power loss – or if a DHCP server goes down how much is it going to cost. If a large organization is decentralized (let all their outlying facilities do what they want), they look to these products to make things more standard. 

Price range for your products? 

QIP: Per IP address basis. $4-5k, can set up as many scopes. Uses Sybase, if customer want  

NetID: Per IP address basis. Sliding scale. $5,000 gets you a NetID 1500 product. Will manage 1500 IP addresses. The more addresses you buy the cheaper the price. Maintenance usually handled separately. 

IN: Based on the number of IP addresses. 7,500 for first 1000, 1500 a year for maintained. Do not require a database, no separate charge. Software only. 

Describe a typical installation: 

QIP: On average, 3 days to install. Depends on import requirements. 

NetID: Average customer buys 20-25k IP addresses. Typically they will install a number of DHCP and DNS servers. Don’t sell on the number of servers (applications), rather they sell based on the number of IP addresses. Don’t pay for redundancy – all part of the base price. Support passive replication of data (using oracle) – however few customers actually use that piece. 

IN: 8 offices. Each DHCP server also hosts a DNS server that also supports DDNS. Every organization is going to implement a different DHCP schema. Primary/Secondary per region. Some have primary secondary per county. Some have only a 1000 users but have numerous locations (as few as 5). 3-5000 nodes typically looking to upgrade. 

Who are some of your customers? 

QIP: financial, manufacturing, healthcare, media, universities. Very strong in financial.  

NetID: Automotive sector, banks, healthcare, universities. 

IN: County of San Dego, Country Companies, and Universities. 

How would you describe your best distribution channels? Web-ecommerce, direct sales, distributors, system integrators, etc? 

Lucent: Direct sale model. Moving over to distributor model similar to Europe. 

NetID: Notel pretty much all go through channels. Few sales are made directly. Most ILECs represent their largest distribution channels. Over 1500 value added resellers. Main skills in the data space. System integrators (IBM, etc.). VARS medium to large. Telco’s all over the map. Looking target managed services play – telcos. 

IN: Enterprise sales mostly done direct. Have two system integrators (one university oriented), and then KPMG 

Future Direction: 

QIP: Evolving IP technologies VPN, wireless, WiFi. Level of authentication required receiving an IP – DHCP authentication possibly required. IPV6.  

NetID: IPV6 next big thing coming. Estimates are always 2 years coming. DNS security also coming down the road. Market space fairly mature. Occasional request to run on Linux. No open source around centralized DHCP solutions. Window 2000, but still a fair amount on NT. Platform does not impact cost – only number of IPs. Haven’t seen any changes in the types of customer. Expanding down – more product awareness.  

With the spin off of the “via” (or AVAYA) group – a very limited set of VARs. Really depends on which distribution channels are left. Avaya has the enterprise solutions. Asend Communications.

Readers who use this information for investment purposes do so at their own risk! Opinions expressed are just that and not based on insider information or information otherwise obtained illegally.

Can Birds-Eye.Net help you or your Company?
Receive your Birds-Eye.Net articles and white papers hot off the presses by adding our RSS feed to your reader.