Lately I was thinking that the way we look at web services is basically delusional. Boom of ideas and functions turned Internet from simple interaction of publishers and readers into powerful media that can swallow and spit any kind and amount of data.
Concept is so large and awesome that it is perceived to have some attributes it doesn’t actually have – persistency and reliability.
Lack of transparency
When working with local computer you know what are you dealing with:
- new or old;
- good condition or failing parts;
- well groomed operating system or outdated wreck.
You can’t see what kind of computer handles your data on other side of Internet (and probably globe as well). Sure some image of giant data center with serious stock photo models engineers walking around pops in mind.
Photo by fred ross lord
Recently online bookmarking service Magnolia went down with complete data loss. Turns out it was operated by staff of one and powered by two servers with few Mac minis on top. Service sure had reached much on such slim resources – the harder was fall.
Lack of long term view
When I registered email account back in 2000 it was for text RPG. Years later I long lost count of services, accounts, password retrieval options, contacts and other stuff that piled up – all tied to very same mail account from free email service. I had no hint of thought that I would even be using that mailbox nine years later.
When subscribing to yet another service we don’t think if it is going to be alive and kicking years later. Clicking that “register account” link doesn’t trigger (at least in mine head) thoughts about:
- what type of my data service will accumulate over time;
- what format it uses for storage;
- what export and import option are available.
Sign up and use, worry later…
Journalspace blog hosting service went down after six years of operation. They had no backup. I don’t know what export features were available but I am very sure most of users assumed that backup is service responsibility and happily used service for years.
Single point of failure
Web services are exactly like offline desktop computer – they are holding some of your data and they may lose it. Web 2.0 aura doesn’t automagically provides services with free pass on technical issues and problem-proof coat.
Click your bookmarks, look at web services and decide what are you going to do if tomorrow those shiny web apps get replaced with dreaded “We are sorry” message.
Lessons to learn
- Data should exist in different places and forms. This post you are reading is a web page. As well as entry in database that will soon become email message in my mailbox and file in my documents with copy on external drive and sync to Dropbox. Does my hosting provider handle backup for me? Yep, that doesn’t keep me from handling my own and hope I won’t ever need them.
- Loss of single copy must leave other copies unaffected. I had once caught one of first viruses that are capable of self-replicating over network (I really miss times when we had to click viruses). Before I knew it my portable software folder was infected and my sync had promptly brought infection to mirror on flash drive. Instead of clean copy I had twice more trouble to handle. It was not really a backup measure but it taught me that mirror is not a viable backup solution. Especially when automated and unchecked.
- No one cares about your data except you. I trust developers of web services to care about their companies, profits, public image, reputation, etc. I don’t trust them to care about my personal data because the person who cares most for that is me and no one else.
My personal experience
After my recent computer meltdown I was glad that it had very little effect on my computing:
- I hadn’t dropped out of contact at all;
- in few hours I had full access to my core applications and data;
- in two days I had hardware replaced and most of software setup restored.
That fortunately was not a data loss situation. But it showed me what I got right in my backup and sync routine and what is missing. And there sure were some things missing that could be very annoying to lose.
It also showed me that not a single online service I use is complete and reliable to handle all of related data out of the box.
Learn to rule the universe or at least SQL
BatteryBar – notebook battery meter done right
Google Sync Services with Nokia smartphone
Passmark’s online collection of benchmark data
Backup best practices guide at dpBestflow.org
Give yourself a backup drill
Revisiting Dropbox and extra space for new users
Review your Google usage at Google Dashboard
Why Google seems evil
That’s funny because I was thinking about the exact same thing this week. There are a lot of services we rely on that are hosted by third parties and there are no guarantees these services will exist forever. Take Feedburner for example: the blogging community relies on it, what if tomorrow the service dies for whatever reason?
The same goes for a lot of other services and that’s why I try not to put the “core” of my activities into another business hands.
Ben.
So, to summarize it all:
1) Look for backup function in services that you use.
2) Never put too much trust in the whole cloud-computing hype.
Drat. Now you make me more paranoid Rarst. I better go reorganize my email and Delicious now.
@Ben
Yeah, there is a lot of fuss about Feedburner lately. I think it’s bit overhyped. Anyway I had my feeds set up with MyBrand so I have control over feed URLs and can leave FeedBurner any time if I want to. :) Again – good and free function that increases your control over stuff but how many bloggers bother to use it?
@MK
That was kinda my goal. :) We don’t think about web services the same way about local apps (install, save files, make backup, move to another app). But we should.