I stayed up late last night and got a message saying the database is doing maintenance when I try to access the Tablo on my shield at around 3am. Is there anyway we can have a configuration option to set the maintenance to take place at a specific time (e.g. 5am)? Different people have different schedules and work shifts, and it will be great to be able to configure that maintenance window based on each person’s individual need. What do you all think?
I would dearly love to know what kind of database is being used anyway in a real-time device! Having worked in factory automation where time was an essential factor, SQL type databases were poison.
Database maintenance? We’re so focused on antenna reception, signal distribution, network management, device streaming…database performance?
Well, you have to store program guides, pointers to recordings, etc. somewhere. Most devices these days use a database - SQLite is very, very common in devices like this.
Doesn’t have to be a SQL database. Can be in other types of data structures more compliant with real-time processing. The move these days has been away from SQL databases for real-time analytics for instance (e.g. Hadoop).
Yes, you’re talking about NoSQL databases that are fine for millions of rows.
Android and iOS apps frequently use something like SQLite because it’s one file, and generally single-user, and fitting for the job. I don’t know what Tablo uses as a data store - but it wouldn’t shock me if it was actually SQLlite. The fact that something is SQL Compliant doesn’t really matter - at the end of the day, you have to manage the data relationships in code if you want to use something flat. For single user products of this type, there’s really no downside to using something like SQLlite - frankly, I’d rather them not have to shred a bunch of flat files or somehow deal with ISAMS - it would be worse than working with a small database.
SQL queries are inefficient for real-time processes. We used BerkeleyDB on real-time factory automation devices and SQL was way behind in query processing. The Tablo has to contend with user processing as well as signal processing simultaneously. When the Tablo CEO made his last announcement, I noticed he mentioned “optimizing queries” as part of the improvement. So there was a database issue. Can that be a problem in the startup delay?
CraigRoyce - since I work with Hadoop, DynaDB, PostGreSQL, and MySQL each and every day - I understand what you’re saying.
I’ll make my final comment this one: good luck installing a Hadoop instance on a Tablo to support a few thousand rows of data that has no query performance concern anyway. Also, good luck dealing with the subtle problems of data getting out of sync that they’re having lately with a NoSQL solution like Hadoop.
A simple metadata database doesn’t need to use the trendiest engine on the block. There’s nothing at all special about Hadoop or other NoSQL solutions - they’re the modern day version of COBOL ISAMS. Relational databases exist for good reason - and for reasonable data sets of even real-time data, they’re perfectly adequate and normally a better tool for the job.
If there’s some notion that database maintenance goes away with something like Hadoop - then you’ve never used it The complaints aren’t performance - the complaint was “why didn’t you let me choose when you did some maintenance routine?” - the maintenance is going to happen, regardless - but it should happen less frequently with no significant performance loss using a relational data store.
That said - I have no idea what Tablo is using.
I never mentioned installing Hadoop on Tablo - I referred to BerkeleyDB used on real-time devices. Hadoop was merely an example of the move away from SQL based processing for real-time analytics.
Right, I understand. Use a key value pair no sql and solve zero of Tablo’s problems - I get it. Pretend I said Berkeley - same issue other than the stack complexity - it still solves zero problems.
Hang on a minute - what happened to the above statement? Don’t you keep your word?
Why should I pretend what you said? You say something, stick with it. You simply misinterpreted my statement and tried to impute something to it that wasn’t there. Where in any post at this forum have you ever seen me backtrack (or use such childish tactics as putting words in somebody else’s mouth)? At least my word is my word. Learn to read…and avoid lying next time…Adam: “I’ll make my final comment this one”…SHEESH
Umm…this is television viewing. It is not a factory floor where we are trying to synchronize robotic laser cutters and welders with a moving conveyor belt. IT IS TELEVISION! What do a few seconds matter?
Forty-five to fifty years ago real time operating systems were much more common in standard commercial business situations. They controlled financial terminals, check sorters, etc.
But over the years CS majors aren’t necessary taught where there is a need for a interrupt driven operating system versus an event driven operating system. And when it’s necessary to have bounded computational dispatch versus time slice dispatch.
Sure thing. But does it really matter for television in my home?
sorry there @CraigRoyce I wasn’t monitoring the forum. I’ll put it this way: You don’t know what you’re talking about and lack the intellectual curiosity to understand platform needs vs. what you know. Tablo’s use of a database has nothing to do with real time processing. You can’t grasp this because you don’t understand the problem, or the complaint. What you do understand is what someone taught you… in one specific application… of a couple specific technologies. If you want to fight about what platform to use, great. But, if you want to fight about different tools being suited for different jobs, you’ll lose most of the time. When your most critical point is to call someone a liar for someone saying they were done with the conversation after you continued to yammer like a 1st year CS student, you bring nothing to the conversation. Sorry I lied and said I made my final comment. My final comment really is to just say you’re not very bright. Have a good day / night / year… whatever.
Great - at least CR and the rest of us will get some sleep tonight!