Eliminate Print Servers: go LANless?
-
@johnhooks said:
Read only via ODBC is perfectly safe.
How? I explained why it can't be. You can't be sure that the end users are using the relationships correctly, so the data is not reliable.
-
@scottalanmiller said:
@johnhooks said:
No that's the whole point. There are a ton of applications that will interface via ODBC and let you pretty much drag and drop.
But we've covered why they can't work here, right? So this point is moot. ODBC can be ruled out as technically impossible. So why does it keep getting mentioned? How is this even considered an option?
It wasn't ruled out at all. You said they may be using NoSQL (or no database at all), which is unlikely as these types of data stores are usually too complex for that.
-
@johnhooks said:
It's fully encrypted from end to end. The same can't be said about the app that's produced if encryption is implemented incorrectly or not at all.
If you can say it about ODBC you can say it about any connection. If you question if this can be enforced on an API, then any logic used there makes ODBC unsafe in the same way. They are equal in being able to enforce encryption. That's not the concern. It's the integrity of the assembled data that ODBC cannot protect.
-
@johnhooks said:
It wasn't ruled out at all. You said they may be using NoSQL (or no database at all), which is unlikely as these types of data stores are usually too complex for that.
Totally ruled out. Using NoSQL is increasingly common and the use of a single database for huge systems like this is almost never going to happen. That some of the systems use NoSQL or something that cannot use ODBC is an extremely real possibility and increasingly so in the future. That it is unlikely to be the sole datasource is very true, but not relevant.
-
@wirestyle22 said:
No I can't. I don't really do anything with mobile devices so I figured I'd ask questions. Sorry I'm not making a ton of sense
VPN = "Adding them to the LAN when they are far away."
That's it. Nothing more.
-
@scottalanmiller said:
@wirestyle22 said:
No I can't. I don't really do anything with mobile devices so I figured I'd ask questions. Sorry I'm not making a ton of sense
VPN = "Adding them to the LAN when they are far away."
That's it. Nothing more.
Yeah but it should make it visible in a way it wasn't before considering we have no MDM system in place. I was wondering how I might use that to my advantage--if its even possible.
-
My unaddressed concerns with the client/server architecture that we keep going round and round on and I've not seen addressed in any way:
- The complexity of combining many data sources when we have no reason to think we have access to the relational information.
- The fact that systems like this often keep relational data that does exist outside of the database and relationships are created and enforced in software, not the database (I'm working with one of those right now, in fact.)
- The fact that ODBC cannot be automated in big systems like this with simple tools because many datasources have to be combined, not just one.
- That the resulting data cannot be provided by a simple "report" as we need to code something to consume many different data sources into one that we massage ourselves and create our own relationships (maybe some tool does this, I don't know of one.)
- The fact that some data sources cannot use ODBC even if relational, but more likely when they are non-relational.
This leaves us with the basic problems: lack of data integrity, lack of resulting product and inaccessibility of data.
-
@wirestyle22 said:
@scottalanmiller said:
@wirestyle22 said:
No I can't. I don't really do anything with mobile devices so I figured I'd ask questions. Sorry I'm not making a ton of sense
VPN = "Adding them to the LAN when they are far away."
That's it. Nothing more.
Yeah but it should make it visible in a way it wasn't before considering we have no MDM system in place. I was wondering how I might use that to my advantage--if its even possible.
VPNs do not make things visible that were not visible before. It just makes them visible when they were not on the local network. If you didn't have the functionality when people were in the office before, adding them to the VPN will not create new functionality.
-
@scottalanmiller said:
@johnhooks said:
It wasn't ruled out at all. You said they may be using NoSQL (or no database at all), which is unlikely as these types of data stores are usually too complex for that.
Totally ruled out. Using NoSQL is increasingly common and the use of a single database for huge systems like this is almost never going to happen. That some of the systems use NoSQL or something that cannot use ODBC is an extremely real possibility and increasingly so in the future. That it is unlikely to be the sole datasource is very true, but not relevant.
It may be increasingly common, but not scenarios like this. The data stores actually storing the data are usually relational.
-
@scottalanmiller said:
My unaddressed concerns with the client/server architecture that we keep going round and round on and I've not seen addressed in any way:
- The complexity of combining many data sources when we have no reason to think we have access to the relational information.
- The fact that systems like this often keep relational data that does exist outside of the database and relationships are created and enforced in software, not the database (I'm working with one of those right now, in fact.)
- The fact that ODBC cannot be automated in big systems like this with simple tools because many datasources have to be combined, not just one.
- That the resulting data cannot be provided by a simple "report" as we need to code something to consume many different data sources into one that we massage ourselves and create our own relationships (maybe some tool does this, I don't know of one.)
- The fact that some data sources cannot use ODBC even if relational, but more likely when they are non-relational.
This leaves us with the basic problems: lack of data integrity, lack of resulting product and inaccessibility of data.
An API is not some magic thing that suddenly provides all of this. An API has to be written to provide this information. It is the very rare API that provides every thing that a user could want to query.
Even if the API was capable of providing all of this mythical access to information, the user would be in the same spot as the person writing the ODBC query. They would be needing to parse down the API returned information to what they want to see.
Is it less complex than learning SQL? Probably a decent amount, but it is far from simple.
-
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
It wasn't ruled out at all. You said they may be using NoSQL (or no database at all), which is unlikely as these types of data stores are usually too complex for that.
Totally ruled out. Using NoSQL is increasingly common and the use of a single database for huge systems like this is almost never going to happen. That some of the systems use NoSQL or something that cannot use ODBC is an extremely real possibility and increasingly so in the future. That it is unlikely to be the sole datasource is very true, but not relevant.
It may be increasingly common, but not scenarios like this. The data stores actually storing the data are usually relational.
That in no way disputes what I said.
-
@JaredBusch said:
@scottalanmiller said:
My unaddressed concerns with the client/server architecture that we keep going round and round on and I've not seen addressed in any way:
- The complexity of combining many data sources when we have no reason to think we have access to the relational information.
- The fact that systems like this often keep relational data that does exist outside of the database and relationships are created and enforced in software, not the database (I'm working with one of those right now, in fact.)
- The fact that ODBC cannot be automated in big systems like this with simple tools because many datasources have to be combined, not just one.
- That the resulting data cannot be provided by a simple "report" as we need to code something to consume many different data sources into one that we massage ourselves and create our own relationships (maybe some tool does this, I don't know of one.)
- The fact that some data sources cannot use ODBC even if relational, but more likely when they are non-relational.
This leaves us with the basic problems: lack of data integrity, lack of resulting product and inaccessibility of data.
An API is not some magic thing that suddenly provides all of this. An API has to be written to provide this information. It is the very rare API that provides every thing that a user could want to query.
Even if the API was capable of providing all of this mythical access to information, the user would be in the same spot as the person writing the ODBC query. They would be needing to parse down the API returned information to what they want to see.
Is it less complex than learning SQL? Probably a decent amount, but it is far from simple.
Thank you, you worded this better than I was able to.
Also what software that's used by any number of people at all does the relations in the software vs in the relational database itself. That defeats the whole purpose of the relational database.
So if we are going to assume bad practices here, why not assume bad practices with the API?
-
@scottalanmiller said:
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
It wasn't ruled out at all. You said they may be using NoSQL (or no database at all), which is unlikely as these types of data stores are usually too complex for that.
Totally ruled out. Using NoSQL is increasingly common and the use of a single database for huge systems like this is almost never going to happen. That some of the systems use NoSQL or something that cannot use ODBC is an extremely real possibility and increasingly so in the future. That it is unlikely to be the sole datasource is very true, but not relevant.
It may be increasingly common, but not scenarios like this. The data stores actually storing the data are usually relational.
That in no way disputes what I said.
It does, because I can pull the info out of the relational database and never need to even know NoSQL exists.
-
@JaredBusch said:
@scottalanmiller said:
My unaddressed concerns with the client/server architecture that we keep going round and round on and I've not seen addressed in any way:
- The complexity of combining many data sources when we have no reason to think we have access to the relational information.
- The fact that systems like this often keep relational data that does exist outside of the database and relationships are created and enforced in software, not the database (I'm working with one of those right now, in fact.)
- The fact that ODBC cannot be automated in big systems like this with simple tools because many datasources have to be combined, not just one.
- That the resulting data cannot be provided by a simple "report" as we need to code something to consume many different data sources into one that we massage ourselves and create our own relationships (maybe some tool does this, I don't know of one.)
- The fact that some data sources cannot use ODBC even if relational, but more likely when they are non-relational.
This leaves us with the basic problems: lack of data integrity, lack of resulting product and inaccessibility of data.
An API is not some magic thing that suddenly provides all of this. An API has to be written to provide this information. It is the very rare API that provides every thing that a user could want to query.
Even if the API was capable of providing all of this mythical access to information, the user would be in the same spot as the person writing the ODBC query. They would be needing to parse down the API returned information to what they want to see.
Is it less complex than learning SQL? Probably a decent amount, but it is far from simple.
I'm not saying that it is simple. I'm just saying that a direct client/server connection leaves you without the normally necessary application layer that assembles and protects the data (in terms of integrity.) It's not magic, but neither is ODBC. The idea with a direct ODBC connection is that a computer can look at a bunch of data and just "know" what it represents but it cannot. The problem is that an API is generally needed here, easy or not, because the other option is generally not even reliably possible (or possibly reliable.)
-
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
It wasn't ruled out at all. You said they may be using NoSQL (or no database at all), which is unlikely as these types of data stores are usually too complex for that.
Totally ruled out. Using NoSQL is increasingly common and the use of a single database for huge systems like this is almost never going to happen. That some of the systems use NoSQL or something that cannot use ODBC is an extremely real possibility and increasingly so in the future. That it is unlikely to be the sole datasource is very true, but not relevant.
It may be increasingly common, but not scenarios like this. The data stores actually storing the data are usually relational.
That in no way disputes what I said.
It does, because I can pull the info out of the relational database and never need to even know NoSQL exists.
Ah, so if most of the data is relational (and supports ODBC) and some doesn't, you can guarantee that the data not available via ODBC can be just dropped or ignored?
-
@johnhooks said:
Also what software that's used by any number of people at all does the relations in the software vs in the relational database itself. That defeats the whole purpose of the relational database.
Welcome to the real world. Tons and tons of applications do this. It's done so commonly that it is a standard problem to look for when fixing in house code. Is it good? Heck no, but it is sometimes required because of the use of multiple data sources and is becoming increasingly common because of the way that data sources exist today and no matter how bad it is, it is common. I run into it constantly when working with companies that develop their own code.
Does Microsoft do this when writing their own products? of course not. But once you have companies making bad APIs, you have the same pool of developers that aren't making good relationships in their databases.
-
@scottalanmiller said:
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
It wasn't ruled out at all. You said they may be using NoSQL (or no database at all), which is unlikely as these types of data stores are usually too complex for that.
Totally ruled out. Using NoSQL is increasingly common and the use of a single database for huge systems like this is almost never going to happen. That some of the systems use NoSQL or something that cannot use ODBC is an extremely real possibility and increasingly so in the future. That it is unlikely to be the sole datasource is very true, but not relevant.
It may be increasingly common, but not scenarios like this. The data stores actually storing the data are usually relational.
That in no way disputes what I said.
It does, because I can pull the info out of the relational database and never need to even know NoSQL exists.
Ah, so if most of the data is relational (and supports ODBC) and some doesn't, you can guarantee that the data not available via ODBC can be just dropped or ignored?
What data are you going to need from a previous doctors visit that isn't permanently stored in the actual storage?
-
@johnhooks said:
So if we are going to assume bad practices here, why not assume bad practices with the API?
I am. And likewise, if we assume good ones, why not assume good ones? I've been pointing out all along that your assumptions were based on a pristine relational database that is fully self describing combined with an API that was hard to use - while possible, it is an unlikely combination and one we would not just assume.
-
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
@scottalanmiller said:
@johnhooks said:
It wasn't ruled out at all. You said they may be using NoSQL (or no database at all), which is unlikely as these types of data stores are usually too complex for that.
Totally ruled out. Using NoSQL is increasingly common and the use of a single database for huge systems like this is almost never going to happen. That some of the systems use NoSQL or something that cannot use ODBC is an extremely real possibility and increasingly so in the future. That it is unlikely to be the sole datasource is very true, but not relevant.
It may be increasingly common, but not scenarios like this. The data stores actually storing the data are usually relational.
That in no way disputes what I said.
It does, because I can pull the info out of the relational database and never need to even know NoSQL exists.
Ah, so if most of the data is relational (and supports ODBC) and some doesn't, you can guarantee that the data not available via ODBC can be just dropped or ignored?
What data are you going to need from a previous doctors visit that isn't permanently stored in the actual storage?
I don't understand this question.
-
Because of the way that patient records are often stored and used, as documents, I'd actually be pretty surprised to not find NoSQL in common usage. A quick search on just one popular NoSQL systems turns up that healthcare is specifically a market that they target and there are lots of resources for using it in exactly this kinds of situations.
