A deadline is an interesting thing. It represents a line in the sand, a point by which a task must be accomplished. Many people — myself included — sometimes treat deadlines as suggestions. And some really are.
Deciding to lose 10 pounds, or taking a vacation in Hawaii are examples of ‘soft’ deadlines. Others are fixed and potentially carry painful consequences: miss a homework assignment deadline, get an “F.” Miss the date for filing your tax return? Get a hefty financial penalty.
Google Search Appliance (GSA) users have an approaching deadline. And missing that deadline will be painful.
The Clock is Ticking for GSA Users
At the end of 2018, your pretty yellow GSA box will become a nice yellow Dell server in need of an operating system.
When Fortune Magazine first announced the news Feb. 4 2016, there were 1060 days left; Martin White and I wrote about it on Feb. 9 when 1055 days remained.
Now, you have less than 1000 days.
As a GSA owner, what have you done so far? What options do you have?
If your content is public facing, Google Site Search might be a fine option.
And Google announced that it is working on a cloud-based replacement for the GSA, apparently in limited beta already. But like me, many people see cloud solutions as potential liabilities. When it comes to confidential corporate information like business financial reports or confidential employee data, an accidental exposure could mean massive fines or even more severe penalties. All sorts of three-letter agencies could become your pals, depending on what leaked.
Know Before You Buy a New Search Platform
You might wonder how hard replacing your existing search platform could be. Dozens of search platforms are out there, both commercial and free open source solutions: how hard can that be?
Not so fast.
I’d support the generalization that all search engines look about the same: you type in a search, and results show up. But different platforms excel in different ways, and without considering your specific requirements, you may end up with an expensive failure.
A few of the points you need to consider:
Content: Not all search platforms support all document formats. Can the search platform use your formats without requiring a specific markup format for documents?
Most — but not all — open source-based projects and products use Apache Tika to index Office and other binary formats. But Tika isn’t perfect, and may not support all the formats you have, especially if you have very old documents. Proprietary search products use their own solutions, or license it from another company.
When a new Office format comes out, how long will it take your search platform to accurately index your content?
Metadata Quality: Consider the validity of your content metadata: do you have useful metadata, or thousands of documents titled “Document 1”?
Repositories: Where do you store documents? Is your content on a file share? Or is it in SharePoint or some other content management system? A database?
Your search solution needs access to the content, so it better work properly with your current — and any future — repositories.
Security Model: Different search platforms implement security filtering differently. Two basic models include "early binding," which indexes the acceptable credential/groups with the content; and "late binding," which links user credentials with the query to ensure proper authorization. For speed and best security, you want early binding. Late binding is easier to implement.
Security Methods: What sort of security do you use? LDAP? Active Directory? Kerberos authentication?
Your search platform has to support the method you use, and you need to verify that it works properly.
Administration: How do you manage the platform? Does it provide a functional UI that manages indexing and search operations? Or does it use configuration files? Can you access search activity reports? How do you tune (or tweak) relevancy?
Open Source vs. Proprietary Costs: Is the entire product open source or proprietary? Know that some products built on free open source are neither free, nor open source, or certain components may be open source and free, but not the entire product.
Software or Appliance: How important is it that the product is available as an "appliance"? Or do you want to use your own servers?
User Expectations: What expectations do all of your users — end users, administrators and IT staff — have? How good are the out-of-the-box results? How hard is it to define "best bets"? How much babysitting does it take to keep it running?
Costs: What are the initial costs, the conversion costs and the ongoing maintenance costs?
Take Action Now
The most important thing is to get started. One thousand days may feel like a long time, but I’ve seen well-executed search evaluations take six months or longer. And once you select your solution, you’ll need time to implement it for your environment.
Start with a brief on-site requirements evaluation that reviews all of the above items, and meet with representative users or power users to listen to their concerns and hear their needs. Some call this a search audit, some call it a search workshop or similar.
Whether or not you have an internal search team, consider outside help. While it may seem a static technology, enterprise search has changed a great deal in the last few years and some excellent open-source — and open-source based — solutions are available.
Find a vendor and platform neutral advisor who understands the enterprise environment and understands enterprise search well. Ask for an outline of how they would approach a two or three day on-site workshop to gather your requirements through meetings with key stakeholders: management, IT, your search team and your users. That engagement should offer enough materials for them to prepare an action plan that includes potential vendors for evaluation, key technology points to address and an evaluation plan.
You should start now. The holiday season of 2017 may see a rush on search platform implementations. And just think — if you're early, you can always try returning your GSA!