Powerlink

October 2010

Frequently Asked Questions: Documentum xPlore Replacement of FAST Instream

Table of Contents
General Questions Timelines and Availability Features Supported Environments Migration Sizing and Performance

Technical Questions

General Questions
Timelines and Availability
Question: Answer: Question: Answer: Question: Answer: Question: Answer: Question: Answer: I have heard talk about Enterprise Search Services (ESS) and Documentum Search Services (DSS). Is Documentum xPlore something new? ESS and DSS were EMC's internal working names for Documentum xPlore. Is this Documentum xPlore a separate product? No. Documentum xPlore is a no-charge feature of the Content Server component of the Documentum Platform In which Documentum version will Documentum xPlore be bundled, and which Documentum versions will have backward compatibility with xPlore? Documentum xPlore will be supported on Documentum 6.5SP2 and above. Documentum 6.7 will be released with xPlore as the primary search function. Do I need an additional license for Documentum xPlore? No, Documentum xPlore is bundled with the Content Server and covered by the same license. When will Documentum xPlore replace FAST? Documentum xPlore will be available for download by Q4. By Documentum 6.7, xPlore will be the only search option. FAST will be supported until December 31, 2011. It seems inconsistent to say FAST is supported on Documentum 6.5 SP3 or Documentum 6.6 and then say FAST is only supported until December, 2011. Can you clarify? Documentum 6.5 SP3 together with FAST or Documentum 6.6 together with FAST are supported to the end of 2011. Any combination of FAST after 2011
1

Question:

Answer:
EMC

Powerlink

October 2010

is unsupported. All support for Fast Instream 4.3.1 feature for Documentum (all versions) terminates as of December 31, 2011. This only applies to the FAST Instream search feature and not to the Content Server overall. Question: Answer: Will any extended support for FAST be offered to customers who cannot complete the transition to xPlore until after December 31, 2011? EMC will support customers using the existing versions of FAST Instream through the end of 2011. After that we will no longer support FAST Instream, and no further extensions to this offering will be added. After that date, we will not be receiving support from Microsoft to address any customer issues with FAST Instream. Support timing and duration for all other aspects of the Content Server and the Documentum platform remain unchanged. When will the documentation on xPlore and the query language xQuery be available? This documentation will be added to Powerlink in Q4. We want to test this product now in our local installation of Documentum 6.5 SP2. Where can we get it? To apply for the software upgrade, which will be available in Q4, log on at http://powerlink.emc.com, click on the Services tab, and then follow the prompts to complete the request. A confirmation email will be sent when the order is processed. This is a no-charge feature of the content server. Where can we see the product working? Does a demo server or video exist? We are in the process of setting up a demo system and will have a booth at Momentum Europe 2010 where you can see and play with xPlore.

Question: Answer: Question: Answer:

Question: Answer:

Features
Question: Answer: Question: Answer: Will xPlore support fuzzy string searching in attributes and full text? Yes, and the behavior is documented, as there are many nuances in behavior (and also configurations that you have at your disposal to change behavior). What does "improved quality of search" mean to me? Optimized for Documentum, xPlore offers improved search functionality drives ease-of-use and increases application performance. Documentum xPlore includes integrated Documentum security for improved response times, as well as drill-down, native-facet computation with security enforcement. For FAST, Documentum recommended not installing this on a VMware machine. Will Documentum now support a VMware image for index server with xPlore? Yes, support for VMware is a major advantage for xPlore that enables costeffective installation options. Do you have any benchmarks comparing FAST Instream with Documentum xPlore?
2

Question:

Answer: Question:

EMC

Powerlink

October 2010

Answer: Question: Answer: Question: Answer: Question: Answer:

We are planning to produce a paper from our internal tests. Does xPlore support as many content formats as FAST? Documentum xPlore will support all the formats that FAST does, but also have newer versions of formats. Where can I see the supported formats? The product documentation for xPlore, posted on Powerlink, lists all formats that are supported. Can multiple Documentum repositories be stored to the same xPlore system? Yes. This is called multi-tenancy. A single xPlore system can support multiple Documentum repositories. These are managed as separate xPlore "domains" and the indexes are logically and physically separate from each domain. Is multi-node supported? Yes, multi-node (also referred to as multi-Instance) is supported out of the box. A node or instance in the xPlore context can be either on the same box or a separate box. Does Documentum xPlore support synonym dictionaries? No, Version 1.0 of xPlore does not support synonym dictionaries. We hope to add support for this in the near future. Does Documentum xPlore support Office 2010 documents? Yes, in fact, Office 2010 formats are the same as Office 2007. Documentum xPlore is able to successfully index and search both formats. Does Documentum xPlore support lemmatization? Yes, we use lemmatization (as opposed to stemming) in Documentum xPlore on both the indexing and search side. The linguistic analysis takes parts of speech into consideration when determining the lemma of the word. What languages does Documentum xPlore support? While our underlying technology can support over 80 languages, we are slowly rolling in certification of the languages. Version 1.0 will support Chinese (simplified), English, and German. Version 1.1 will add support for French, Italian, Spanish, Japanese, and Korean. I have content in the languages slated for the 1.1 release. When can I expect that release? xPlore 1.1 will be bundled into Documentum 6.7 release, which is targeted for Q1 2011. Is there any plan to support Portuguese? Yes, it is on our list, but there is no firm roadmap yet. Brazilian Portuguese is in the next wave of languages we will add after our 6.7 release.
3

Question: Answer:

Question: Answer: Question: Answer: Question: Answer:

Question: Answer:

Question: Answer: Question: Answer:

EMC

Powerlink

October 2010

Question: Answer:

What is the future Roadmap for Documentum xPlore? Roadmap information is included in presentations that will be available on Powerlink in Q4.

Technical Questions
Supported Environments
Question: Answer: Which Operating Systems support xPlore? Documentum xPlore will run on Linux and Windows (physical or virtual), but the Documentum Content Server can be run in all the usually supported OS (Windows, Linux, Solaris, HP, and AIX). Do we intend to support platforms other than Linux and Windows? At present, there are no plans to support xPlore itself on other platforms. Of course, your Content Server can be on any of the other platforms, but xPlore itself must be on Windows or Linux. What are the OS versions supported for Windows? Documentum xPlore supports both Windows 2003 and Windows 2008. Both 32-bit and 64-bit. Refer to the Certification Requirements Document (CRD) on Powerlink for further information. Can I still use Documentum xPlore if my Content Server is on HPUX? Yes, although Documentum xPlore is only offered on Linux and Windows, it is compatible with Content Server running on any platform (Window, Linux, HPUX, AIX, and Solaris) Does xPlore support 64-bit Operating Systems? Yes, xPlore supports both Windows and Linux 64-bit versions. Refer to the CRD for the specific OS versions that are supported, as they change with each release. The core xPlore is native 64-bit, but one componentContent Processing Service (CPS)is not native 64-bit in version 1.0 due to third-party dependencies. In the 1.1 release, both the core xPlore and CPS will be native 64-bit. Will 64-bit Content Servers support FAST for 6.6? No, although FAST can be run on a 64-bit OS, it is a 32-bit application. There are no plans to provide a 64-bit version of FAST. Will xPlore work properly with Branch Office Caching Services (BOCS)? Documentum xPlore is compatible with BOCS, but xPlore always retrieves data directly from the content server. Will EMC provide any best practices guidance for the transition to xPlore? Yes, a Documentum Best Practices Guide will be posted on Powerlink in Q4.

Question: Answer:

Question: Answer:

Question: Answer:

Question: Answer:

Question: Answer: Question: Answer: Question: Answer:

EMC

Powerlink

October 2010

Question: Answer:

Does Documentum xPlore support high availability? Yes, xPlore supports multiple types of high availability, including active/passive with clusters, active/active with shared data, n+1 Spare Node HA, and others. Is xPlore supported for CenterStage and Federated Search Services? Yes. CenterStage 1.0 and later support xPlore, but CenterStage 1.1 and newer will leverage new capabilities of xPlore. Federated Search is also compatible with xPlore. Is Documentum xPlore a significantly different architecture than FAST, or is it just a simple swap of one search engine for another? We use the same plug-in architecture on the content server side, but xPlore is an entirely new component. Can I have multiple instances of xPlore running on the same server? Yes, you can have multiple instances of xPlore running on the same box, and even have a single instance of xPlore support multiple repositories. Are storage area networks (SANs) supported for xPlore? Yes. SANs are supported. Are NAS devices supported for xPlore? Yes, they are supported, but we recommend performance testing such a configuration prior to going into production to ensure that the NAS has been set up properly. Are there any restrictions on support for SAN or NAS storage? No, xPlore supports both SAN and NAS. The performance on NAS will depend on the speed/configuration of the NAS but we have tested on NAS and will provide guidance on how to ensure xPlore works well on NAS. What about integration with other search engines? How will xPlore work with those? xPlore does not provide the integration to other search engines. For this functionality customers should leverage Federated Search Services, whose purpose is to integrate with other search engines Are there any restrictions in xPlore with respect to database? No, xPlore has no database dependency. Can customers install search on the same hardware as the Content Servers in a UNIX shop? Documentum xPlore runs on Linux and Windows, and on those operating systems you can run both the content server and xPlore on the same server, given that it is sized appropriately.

Question: Answer:

Question: Answer: Question: Answer: Question: Answer: Question: Answer:

Question: Answer:

Question: Answer:

Question: Answer: Question: Answer:

EMC

Powerlink

October 2010

Migration
Question: Answer: Will my application need to change when I migrate to xPlore? xPlore is compatible with Documentum 6.5 SP2 and onwards. Applications do not have to change to leverage it. The only exception is that the deprecated legacy use of SEARCH TOPIC (with the Verity query language) is no longer supported. This functionality (for XML zone search) has been replaced with xQuery. Why should I deploy the 1.0 product? We understand the importance of quality and stability regarding such a fundamental capability of the platform. That is why we insisted that Documentum xPlore successfully meet one of our most rigorous and comprehensive customer validation processes. We performed an extended six month beta and as of October are in our controlled release program with 25 multi-tier customers who are committed to full production. Will it require a complete re-index of all content, or can we somehow import the relevant index data which would prevent that? Yes, a one-time re-index is required. Do we expect it to be an easy or difficult migration? Documentum xPlore has been designed to make migration from FAST easy for existing customers. Aside from improved deployability, our search feature supports the ability to run in parallel with the FAST search feature to allow a rolling upgrade from FAST to the new search. You also have the option of engaging EMC's Professional Services organization or trusted EMC partner organizations to accelerate implementation and minimize any potential disruption. Will xPlore be available as separate installer if we want to use it with Documentum 6.5 SP2? Yes, xPlore will be available as a separate installer and will include the core Documentum xPlore and an updated Index Agent. Are partners and EMC Professional Services planning offerings to assist customer transitions to xPlore? Yes. EMC Professional Services plan to offer both a standard and comprehensive migration offering. Many of our partners have also shown interest in providing additional service offerings. How long will a migration take? This depends on a few factors, including:
EMC

Question: Answer:

Question: Answer: Question: Answer:

Question: Answer: Question: Answer:

Question: Answer:

The size of the corpus (both number of documents and size of documents). The hardware involved (network, hosts, disk subsystem, etc.).

Powerlink

October 2010

Question: Answer:

During the rolling upgrade, can I run both FAST and xPlore on the same SAN? This can be done, but it is important to ensure that the SAN volume has sufficient I/O capacity (not just disk space capacity) to handle the combined volume. During the rolling upgrade, can I run both FAST and xPlore on the same host? This is a risky approach to migration, unless the repository is very small. The machine could actually support both search engines if there are only a few users. In general, it is best to have xPlore run on a separate server.

Question: Answer:

Sizing and Performance

Question: Answer: Question: Answer: Which disk I/O architecture will perform better, NAS or SAN? For similar bandwidths, numbers of drives, and cache, a SAN will typically perform better than a NAS device for xPlore. What is the overhead when running xPlore on a VMware environment? Internal tests have shown a 15 percent overhead but this could vary based on the number of cores and the number of VMs allocated on the physical hosts, among other factors. What are some of the factors I should take into consideration when picking an underlying disk I/O subsystem? There are some obvious ones like performance and cost, but also the high availability (HA) strategy and the disaster recovery (DR) strategy you intend to deploy. For example, choosing just SAN storage rules out the N+1 sparing strategy for HA. Other options in this particular case include active/passive clusters (like Microsoft Cluster server for Windows) and dual system active/active. See the xPlore deployment guide on Powerlink for more details. Question: Answer: The migration is going slower than I expected and I have plenty of CPU available. What could be going wrong? The problem could be that the Storage Area Network is slow so the feeding of data from Documentum is taking longer than anticipated. Another possibility is that the network between the content server and xPlore is slower than expected. Finally, another common problem is that the disk used to store the xPlore data is slow. Should I locate my xPlore data on the same NAS volumes that hold my Documentum content? If one can ensure that the NAS volume has plenty of bandwidth, low latency, and the ability to handle large numbers of random I/Os, then that architecture may work well for xPlore. But if this NAS volume was designed for infrequent, fairly linear I/O, then it may not be suitable. Is VMware supported for xPlore?
7

Question: Answer:

Question: Answer:

Question:
EMC

Powerlink

October 2010

Answer: Question: Answer:

Yes, unlike with FAST, VMware is fully supported for xPlore. I'm considering deploying on VMware. Are there any tips I need to be aware of? Yes. In many cases IT groups will allocate VMs with a very small initial size (because it is easy to add more resources). Don't assume that the size they pick is correct. It could easily be undersized. Also, note that they could allocate the VM's on a host that is highly consumed. This can also lead to intermittent performance issues. Is there a sizing spreadsheet for xPlore? Yes. It will be available on Powerlink in 4Q. I'm considering deploying the disk to a SAN. Are there any tips I need to be aware of? Not all SANs have the same capacity and performance. xPlore will have better performance on higher capacity SANs than lower capacity ones. In this context, "capacity" is the ability to perform random I/Os. Note that in many cases a SAN volume might be sharing the same set of drives that other volumes are using. It is good to set the expectation to the SAN administrator that sharing xPlore's disk capacity with other intensive disk applications could lead to slower than desired response time and throughput. I have several repositories and their versions are Documentum 6.5 SP2, Documentum 6.5 SP3, and Documentum 6.6. Can I have all of these be served by the same xPlore system? Yes. xPlore supports multiple versions of Documentum on the same host. How much memory should I allocate for xPlore? This depends mostly on the size of the repository, the query rate, the speed of the disk I/O subsystem, and the expectations of response time. A lot of users typically require more memory. One can compensate for an overworked or slow disk I/O subsystem by allocating more memory for the OS buffer cache. See the sizing spreadsheet that will be available in Q4 for more details. How much disk space do I need to allocate for xPlore? This depends on the size of the system (number of documents) and a few other factors. Typically, xPlore implementations will consume less disk space than FAST deployments. Hence xPlore should consume less disk space than FAST for equally sized repositories. So it would be safe to allocate the same space for xPlore as was allocated for FAST (assuming, of course, that the FAST environment had sufficient space allocation). See the sizing spreadsheet that will be available in Q4 for more details. Does my choice of xPlore high availability architecture affect my sizing? Yes. The different sizing options have different resource needs. For example, the active/active dual system approach requires duplicate hosts and duplicate disks. The active/passive cluster typically only requires dual hosts because the
8

Question: Answer: Question: Answer:

Question:

Answer: Question: Answer:

Question: Answer:

Question: Answer:

EMC

Powerlink

October 2010

disk is shared between them (only active for the active server). The N+1 sparing strategy requires no duplicate disks but does require some number of duplicate servers. See the deployment guide that will be available on Powerlink in Q4 for more details. Question: Answer: Question: Answer: Do customers typically size for back file migration or day-forward ingestion? Typically, customers size for day-forward processing, which includes query as well as ingestion. Can xPlore and the Documentum Index Agent be deployed on the same host? Given a Linux and Windows environment, yes. It is just necessary to ensure that there is sufficient memory, CPU, and disk to accommodate both applications. However, what we are seeing so far in our controlled release is that a majority of customers are using VMware and deploying xPlore on its own VM. Can my choice of a disaster recovery strategy affect my sizing? It can. For single-site DR, the choice of the recovery time objective (RTO) could affect how often a system is backed up and how much extra storage is required. For example, applications that are not all that mission-critical might opt out of performing a backup and re-feed the data from Documentum in case of a disaster. In this case the RTO could be measured in days. Applications that are more mission-critical can take advantage of xPlore's hot backup capability for a shorter RTO (hours or minutes). Will EMC provide any infrastructure planning guides so that customers can purchase proper infrastructure or re-purpose existing infrastructure for xPlore? Yes, a sizing spreadsheet and presentation will be available in Q4. Will EMC provide any benchmark materials for customer consumption so that customers can plan for the time to index with xPlore? Yes, a benchmark will be posted on Powerlink in Q4. In general, will xPlore require smaller or larger hardware requirements than FAST did for a the same database? The requirements will be smaller in index size and smaller in overall resource requirements. And if you consider a VMware deployment, actual hardware can be minimized. One of the problems with FAST was the large disk space requirements for FAST while creating an index. Does xPlore have similar issues, or has this been handled? Documentum xPlore uses much less transient disk space during indexing than FAST, so this should be much better. We've done extensive testing in this area to tune and ensure xPlore behaves well.

Question: Answer:

Question: Answer: Question: Answer: Question: Answer:

Question:

Answer:

EMC

Powerlink

October 2010

Question: Answer:

Any idea when the sizing tools will be updated to use the most efficient xPlore calculations? The updated sizing tools are being validated in our controlled release and should be available Q4.

EMC

10