Planning

The migration from Site Server 3 to SharePoint Portal Server for intranet search at Microsoft included the following stages: Planning, Analysis and Design, Deployment, and Management.

Identifying Deployment Goals

In addition to running the enterprise IT utility, ITG plays a strategic role as one of Microsoft's early adopters, testing and deploying Microsoft software before customer release. All ITG early adoption efforts must show tangible business benefits to Microsoft beyond testing for scale and load in a real-world production environment. This was true for the SharePoint Portal Server beta deployments.

Among other benefits and services, this deployment extends the "Microsoft software as a service" model to continue to provide:

• A customer-specific, service-level agreement for each portal owner that defined the service and clearly stated the procedures for support and maintenance over time.
• Search across multiple (even disparate) content sets.
• Better performance and more timely and relevant results.
• The inclusion of existing content and additional content in the index.

The project team established one key metric to measure their success. The team had to ensure that the system handled the stress of crawling about 6 million documents in a time frame that matched their existing results. The existing enterprise search solution included only about 3 million documents in an index. The team also planned to add additional intranet content to the indexes. In addition, ITG required additional room for the growth of content over time.

To verify that SharePoint Portal Server would handle the same load as Site Server 3, the team ran both products in parallel for 30 days before retiring the Site Server 3 solution.

Establishing a Project Timeline

ITG began planning in the summer of 2000 to test SharePoint Portal Server as an enterprise index and search technology through all interim releases, including Beta 1, Beta 2, Release Candidates, and the final release-to-manufacturing (RTM) version.

The team divided the project into the following four phases:

Planning

• Establish the team.
• Collect information on the current environment.
• Develop a project plan.

Analysis and Design

• Create the architecture and select the hardware.
• Review and redefine the catalogs.

Deploying

• Install the hardware and software.
• Configure servers running SharePoint Portal Server.
• Create workspaces.
• Set up content sources and site rules.
• Complete property mapping from custom document properties to the SharePoint Portal Server schema.
• Modify Active Server Pages (ASPs) for searching and for returning results.
• Test crawling.
• Test searching.
• Operate Site Server 3 and SharePoint Portal Server in parallel.

Managing

• Make the transition to production.
• Manage operations and perform maintenance.

The team spent about nine months on this effort from beginning to end, working part-time. From midsummer when the team was formed until the end of the year 2000, the team spent most of its time testing the index and search capabilities of SharePoint Portal Server in beta and optimizing for the goal of 6 million documents, as shown in Figure 27.1.

The migration to production began in early January 2001 with development of the search page and completion of the final tests of crawling. In mid-February, ITG set up the parallel environment. Before RTM in mid-March, SharePoint Portal Server replaced Site Server 3 for search queries on the primary corporate portal, called MSWeb, and the Product Group Portal. After RTM, ITG began converting all major portals at Microsoft to SharePoint Portal Server for search. When this process is complete, Microsoft will retire the Site Server 3 solution throughout the corporate intranet.

Figure 27.1. Project and development timeline

Based on its experience, the team estimates that a typical enterprise customer migration of similar scale might take approximately three months, as illustrated in the following table.

Table 27.1   Typical Enterprise Project Timeline

	Month 1	Month 2	Month 3
1. Planning	1 week
2. Catalog review (optional) Note: can parallel activities 3 and 4	1–4 weeks
3. Hardware installation and setup	1 week
4. Configuration of servers and workspaces, and setup of site rules	1 week
5. Test of crawling operations		2 weeks
6. Modify existing custom ASP pages		1 week
7. Test of search page		1 week
8. Parallel operations			2–4 weeks
Ongoing catalog and index review

Collecting Information

The next part of the planning process included collecting critical information about the existing environment, including several critical components.

Hardware Specifications

Hardware specifications for both crawl and search servers:

• Specify similar hardware for test comparison.
• Comply with upgrades in accordance with ITG hardware standards.

Architecture Diagrams

Architecture diagrams, indicating:

• Hardware, network, propagation paths

Catalog Information

List of all Site Server 3 catalogs, including:

• The server on which the catalogs are stored
• Who owns the catalogs
• How frequently the catalogs are crawled
• What start addresses and site rules are contained in the catalogs
• Whether complex URLs are enabled

Key metrics for each individual catalog and across all catalogs:

• For index (per catalog): number of site rules, number of documents, catalog size, time to crawl, and propagation time
• For search (total and per catalog): number of queries per month and at peak load

Network Environment

Network factors, including:

• Networking protocols
• Firewall configuration

The project team examined the existing network environment for possible factors that would affect deployment, but determined that they did not need to make any configuration changes.

User Environment

Unique environmental factors including:

• The Microsoft corpnet spans the world. Consequently, SharePoint Portal Server must crawl documents in multiple languages for inclusion in the content indexes and must allow users to submit queries in multiple languages. SharePoint Portal Server allows users to submit queries in English, French, Italian, German, Swedish, Spanish, Dutch, Japanese, Chinese Simplified, Chinese Traditional, Korean, and Thai.
• Corpnet is in use worldwide 24 hours a day, 7 days a week, 365 days a year.
• Security is enforced per document at the file share level.