Heartbleed has already been confirmed at the initial attack vector for the breach of a large healthcare system that stole 4.5M patient identities. Given the difficulty tracing Heartbleed attacks, it's likely other systems were compromised in the same way.
Index only scans were added to postgresql (some caveats) in 9.2. The optimizer is cost/statistics based, though perhaps marginally less mature.
What I miss are strong partitioning support, implicit query parallelism, incremental backups, clustering (RAC), and materialized views. Most / all of these features matter primarily for reporting / analytic workloads.
PostgreSQL is a superb database, and dramatically easier to work with and manage than Oracle on a day to day basis. For transactional workloads at anything but the largest scale, it's excellent. On reporting and analytic workloads, it hits the wall much earlier but is still a good option for many needs.
Google Health and Microsoft HealthVault's personal health records (PHRs) are well known in health IT circles, but even among the health IT and healthcare informatics professionals I work with, uptake has been very shallow. There have been connected PHR-enabled sensors available for weight, blood pressure, blood glucose, and many other biometrics for some time, but again, very little interest in flowing this data into stand alone PHRs.
Stand alone PHRs aren't the only way to facilitate doctor-patient interaction. Many leading electronic medical records systems (EMRs) offer integrated personal health records - the disadvantage being that these records only show the data from one provider or health care system. Health Information Exchanges (HIEs) are rapidly springing up across the country to facilitate provider to provider data integration and provide a compelling model for direct patient participation in their care.
Personally, I've tracked these services for years but I've never bothered to create an account. Entering my information manually is tedious, and the standards and integration between EMRs and stand-along PHRs is emerging at best. If I had a fully populated PHR, it's not clear what value I'd really get out of it. My main provider already has most of my information and can source information directly from other practices when needed. Doctors are culturally suspicious of patient submitted data, as they have concerns about amateur self-diagnosis and drug-seeking patients.
The way Google is winding this down increases my trust in their other services. Google announced their plan to shutter Google Health a year and a half before the final shutdown date. They're offering multiple data export and migration options, including instructions and support to migrate to their largest competitor, HealthVault. I've had significantly worse experiences with migration / upgrade of many paid services / software - I'm looking at you Intuit.
Oracle and DB2 both support the SQL/XML standard and provide quite a bit of functionality for native handling of XML. Both can store structured / compressed representations in a native XML type (with or without a predefined schema) and use XPath-based indexes for efficient query execution.
Wonderful stuff, and one of the few features I really miss back in the PostgreSQL world.
From where I stand, either IBM or Oracle were good matches.
Regarding Microsoft - as far as I know, the majority of Oracle's database revenue is coming in on Solaris/sparc and Linux/x86 platforms. Oracle was already squarely in the Java / Linux camp before the acquisition - their applications and middleware stacks are almost entirely Java (or moving there), and they have their own Linux distribution.
The Java acquisition is an imporant defensive move - there are too many free languages and tools out there to make a significant revenue directly from selling languages. With Oracle and IBM both distributing JVMs, Java has a high probability of holding on to its vibrant developer community. I suspect Oracle had concerns about relying on IBM (a very direct competitor) for Java (a foundation technology for Oracle products). The cost of running Java is nothing compared to the costs Oracle would experience if Java either faltered or was tilted against them. Oracle actually built their own JVM for use within the database years ago, though it's essentially dormant now.
The same likely goes for Sun's commercial middleware offerings - Oracle is likely interested in incorporating the best pieces of these into their own stack (see BEA, PeopleSoft, JD Edwards, etc.).
MySQL may be different story - Oracle doesn't seem to have any strategic need for it and they already sell / give away smaller editions of the Oracle DB. I'm personally expecting Oracle's MySQL distribution to wither on the vine but some of the open source forks have a good shot at thriving.
Sparc is likely to continue dying off slowly, whether in Oracle's hands or as a spun off property. There are still plenty of companies happy to upgrade legacy boxes at this point and there's non-trivial revenue that can be derived from them, but the high end systems market will likely consolidate into Power and Itanium, as well as being absorbed into x86. Sparc doesn't have a champion any more and the larger price / architectural trends don't look good for it.
Solaris is an interesting piece of technology and it's less clear what's likely to happen to it. Oracle may see strategic benefits to distributing it themselves as a favored , or they may take some of the technologies (ZFS, scheduler, etc.) and port them to Linux.
OpenOffice and NetBeans are also less clear. There doesn't seem to be significant revenue or strategic potential for either unless Oracle wants to go up against MS Office. Oracle already has JDeveloper in the IDE space and it seems more likely they'd build on Eclipse if there's interest in a new platform.
A centimillion only buys one decitesla. I suspect the editors were referring to hectomillionaries, though with the recent gyrations in the market some of them may have dropped an exponent or two.
For bikes, aluminum frames are certainly stiffer, but they're substantially less shock absorbing.
One of the reasons carbon fiber is used is the ability to choose different properties on different axes. Many cyclists want a frame that absorbs road vibration (longitudinally flexible) while being as stiff as possible laterally to transmit pedaling force efficiently and maneuver aggressively.
I've always thought that in the context of governments, big numbers should be reported on / discussed in per-capita dollars.
$60Bn / year is $200 per capita. All things being equal (they're not), I spend two weeks worth of grocery money on national intelligence every year.
Per capita numbers are great because they work at all levels of government. If your city proposes spending $10m to repair a few local roads, you can evaluate the $50 per capita that represents against projects at other levels of government.
Hey, now there's a service I'd like to see from Google: search web sites from the future. That would put an end to all of this tiring speculation on what new service they'll think up next.
Come to think of it, they could incorporate the technology into other parts of the side. Why present a list of results when you can search the future logs to find out which result I'm going to click and take me straight there.
Except that in the case of what I observed, Prof. Hawley would literally drop an in-progress project the moment the media buzz died down.
This is different from popular researchers such as Carl Sagan and Steven Hawking, who routinely give/gave simplified glimpses of their research to the public, but certainly haven't driven their research based on how much media exposure it's likely to generate.
I knew and interacted with Michael Hawley lightly for a year (temporary advisor at MIT).
From my experience, he was constantly chasing whatever research line was most likely to get him in the media while neglecting projects that seemed to have more research merit but less potential for media attention.
Actually, the force on both objects is necessarily equal and opposite in direction. You are correct that F=ma plays a role though.
If m(cannon) = 100 and m(projectile) = 1, and the forces are equal, let's say F=1, then the acceleration a(cannon) will be 1/100 =.01, but the acceleration of the projectile will be 1/1 = 1.
I say we transfer the program to the deparement of education on the grounds that it'll make for some good physics textbook examples.
I'm looking across my desk at my: - 1 year old Dell 1.8 GHz P4 - 3 year old Apple Powerbook 400 Mhz G4
(similar RAM on both)
The G4 takes about twice as long to compile a large java project (5000 classes) as the Dell, which isn't bad for a 3 year old machine.
On the other hand, the Dell's DVD drive is broken, the latch for the battery keeps coming loose, and the machine gets hot and runs fan while it's sleeping.
The powerbook has intermittent problems with a clicking / scraping sound while watching DVDs. That's it. Battery life is still excelent, and the OS keeps getting faster with each point release.
Oddly enough, my mac (OS X) runs all of those forms/swing apps I've tried without a complaint even though it's obviously not running JInitiator and isn't certified to work. Same goes for their JDeveloper IDE.
Given the apparent lack of JVM incompatability, I'd imagine they'll just do away with JInitiator.
One good reason you might want the database to have some say in the caching is to prevent cache pollution.
Oracle, for example, implements a split cache with hot and cold halves. Small reads get added to the head cache, while major reads (such as full table / index scans over a certain threshold) are added half way down the LRU stack in the "cold" half.
The idea here is that full scan type activity is less likely to provide cache hits on future queries. Also- full scan blocks come at a much lower cost per block than single, random access blocks, and therefore represent a lower value / storage ratio.
This is the kind of guess that's difficult for an operating system to make.
There's no special "compatibility" mode that I know of, except for the fact that it's a UNIX, and therefore very similar to Linux.
Many apps can be ported with a simple compile (most of the GNU stuff just works). Oracle ported their entire database (although it's still considered beta) with minimal effort.
In a commercial setting though, testing, documentation, and support will always cost real money.
I'm doing exactly what you're talking about using CVS.
Everything I care about lives in huge CVS project, currently hosted on a laptop. In my case, this comes to 7GB, mostly in my photo gallery.
For servers (Solaris) at work, everything shares one NFS mounted home directory, which has a checkout of everything but my photos (due to quota limits). My personal work machine (Win2k, yuck) has its own complete checkout, with a few partial checkouts to put configuration files where different programs want them. My home machine (OS X) has a complete checkout.
This setup works for everything except bookmarks, where I find myself using IE on Windows, Netscape in Unix, and Safari on OS X. For bookmarks, I've moved to the "My Yahoo" service, which keeps everything online and accessable with one click from my home page.
By the way, I'd strongly recommend configuring CVS to default to binaries unless told otherwise for this kind of config. It's a nasty surprise to realize that CVS just mangled all of your.xyz binary files trying to do linefeed conversion, just because you forgot to tell it otherwise.
My 1.5 year old TiBook 400 still gets 3 hours or so with full brightness. I generally only discharge down to 50% or so, but it's certainly been down to zero several dozen times.
This is much better than the Dell laptop I had in the past, which spent almost all of it's time plugged in but was only getting 30-45 minutes after a year of use. It also ran the fan intermittently while in standby, which never made much sense.
More than that, I'm always amazed at terrible coding pain non-CS people frequently subject themselves to by not using abstraction, source control, simple comments, or even descriptive names for variables and functions. This is in spite of the fact that many of these people do nothing but write code all day long.
Slashdot Mirror
Heartbleed has already been confirmed at the initial attack vector for the breach of a large healthcare system that stole 4.5M patient identities. Given the difficulty tracing Heartbleed attacks, it's likely other systems were compromised in the same way.
Index only scans were added to postgresql (some caveats) in 9.2. The optimizer is cost/statistics based, though perhaps marginally less mature.
What I miss are strong partitioning support, implicit query parallelism, incremental backups, clustering (RAC), and materialized views. Most / all of these features matter primarily for reporting / analytic workloads.
PostgreSQL is a superb database, and dramatically easier to work with and manage than Oracle on a day to day basis. For transactional workloads at anything but the largest scale, it's excellent. On reporting and analytic workloads, it hits the wall much earlier but is still a good option for many needs.
Google Health and Microsoft HealthVault's personal health records (PHRs) are well known in health IT circles, but even among the health IT and healthcare informatics professionals I work with, uptake has been very shallow. There have been connected PHR-enabled sensors available for weight, blood pressure, blood glucose, and many other biometrics for some time, but again, very little interest in flowing this data into stand alone PHRs.
Stand alone PHRs aren't the only way to facilitate doctor-patient interaction. Many leading electronic medical records systems (EMRs) offer integrated personal health records - the disadvantage being that these records only show the data from one provider or health care system. Health Information Exchanges (HIEs) are rapidly springing up across the country to facilitate provider to provider data integration and provide a compelling model for direct patient participation in their care.
Personally, I've tracked these services for years but I've never bothered to create an account. Entering my information manually is tedious, and the standards and integration between EMRs and stand-along PHRs is emerging at best. If I had a fully populated PHR, it's not clear what value I'd really get out of it. My main provider already has most of my information and can source information directly from other practices when needed. Doctors are culturally suspicious of patient submitted data, as they have concerns about amateur self-diagnosis and drug-seeking patients.
The way Google is winding this down increases my trust in their other services. Google announced their plan to shutter Google Health a year and a half before the final shutdown date. They're offering multiple data export and migration options, including instructions and support to migrate to their largest competitor, HealthVault. I've had significantly worse experiences with migration / upgrade of many paid services / software - I'm looking at you Intuit.
Oracle and DB2 both support the SQL/XML standard and provide quite a bit of functionality for native handling of XML. Both can store structured / compressed representations in a native XML type (with or without a predefined schema) and use XPath-based indexes for efficient query execution.
Wonderful stuff, and one of the few features I really miss back in the PostgreSQL world.
From where I stand, either IBM or Oracle were good matches.
Regarding Microsoft - as far as I know, the majority of Oracle's database revenue is coming in on Solaris/sparc and Linux/x86 platforms. Oracle was already squarely in the Java / Linux camp before the acquisition - their applications and middleware stacks are almost entirely Java (or moving there), and they have their own Linux distribution.
The Java acquisition is an imporant defensive move - there are too many free languages and tools out there to make a significant revenue directly from selling languages. With Oracle and IBM both distributing JVMs, Java has a high probability of holding on to its vibrant developer community. I suspect Oracle had concerns about relying on IBM (a very direct competitor) for Java (a foundation technology for Oracle products). The cost of running Java is nothing compared to the costs Oracle would experience if Java either faltered or was tilted against them. Oracle actually built their own JVM for use within the database years ago, though it's essentially dormant now.
The same likely goes for Sun's commercial middleware offerings - Oracle is likely interested in incorporating the best pieces of these into their own stack (see BEA, PeopleSoft, JD Edwards, etc.).
MySQL may be different story - Oracle doesn't seem to have any strategic need for it and they already sell / give away smaller editions of the Oracle DB. I'm personally expecting Oracle's MySQL distribution to wither on the vine but some of the open source forks have a good shot at thriving.
Sparc is likely to continue dying off slowly, whether in Oracle's hands or as a spun off property. There are still plenty of companies happy to upgrade legacy boxes at this point and there's non-trivial revenue that can be derived from them, but the high end systems market will likely consolidate into Power and Itanium, as well as being absorbed into x86. Sparc doesn't have a champion any more and the larger price / architectural trends don't look good for it.
Solaris is an interesting piece of technology and it's less clear what's likely to happen to it. Oracle may see strategic benefits to distributing it themselves as a favored , or they may take some of the technologies (ZFS, scheduler, etc.) and port them to Linux.
OpenOffice and NetBeans are also less clear. There doesn't seem to be significant revenue or strategic potential for either unless Oracle wants to go up against MS Office. Oracle already has JDeveloper in the IDE space and it seems more likely they'd build on Eclipse if there's interest in a new platform.
A centimillion only buys one decitesla. I suspect the editors were referring to hectomillionaries, though with the recent gyrations in the market some of them may have dropped an exponent or two.
For bikes, aluminum frames are certainly stiffer, but they're substantially less shock absorbing.
One of the reasons carbon fiber is used is the ability to choose different properties on different axes. Many cyclists want a frame that absorbs road vibration (longitudinally flexible) while being as stiff as possible laterally to transmit pedaling force efficiently and maneuver aggressively.
I've always thought that in the context of governments, big numbers should be reported on / discussed in per-capita dollars.
$60Bn / year is $200 per capita. All things being equal (they're not), I spend two weeks worth of grocery money on national intelligence every year.
Per capita numbers are great because they work at all levels of government. If your city proposes spending $10m to repair a few local roads, you can evaluate the $50 per capita that represents against projects at other levels of government.
Hey, now there's a service I'd like to see from Google: search web sites from the future. That would put an end to all of this tiring speculation on what new service they'll think up next.
Come to think of it, they could incorporate the technology into other parts of the side. Why present a list of results when you can search the future logs to find out which result I'm going to click and take me straight there.
HL7 v2 was a problem precisely because the standard was imprecise and full of "other" fields (named Z-segments if I remember correctly).
V3 is a different beast entirely. I'd have a hard time naming another spec with the same level of rigor and documentation.
XEmacs is easy to install via fink- "fink install xemacs" and go grab a long lunch while it compiles for 2 hours...
It not a GUI-native application, so you'll need to run it in X, but it's completely functional. I've been using it for 3 years this way.
(limb breaks)
That's exactly the idea of a consistent backup, or of any non-blocking read consistent query.
The implementation varies from DB to DB (Postgres keeps Multiple Versions of Concurrently updated data for as long as necessary, MVCC).
Oracle has it. IBM has it. Postgres has it. MySQL doesn't.
Um, PostgreSQL anyone?
Nope.
Except that in the case of what I observed, Prof. Hawley would literally drop an in-progress project the moment the media buzz died down.
This is different from popular researchers such as Carl Sagan and Steven Hawking, who routinely give/gave simplified glimpses of their research to the public, but certainly haven't driven their research based on how much media exposure it's likely to generate.
I knew and interacted with Michael Hawley lightly for a year (temporary advisor at MIT).
From my experience, he was constantly chasing whatever research line was most likely to get him in the media while neglecting projects that seemed to have more research merit but less potential for media attention.
Sigh, slashdot physics... :)
.01, but the acceleration of the projectile will be 1/1 = 1.
Actually, the force on both objects is necessarily equal and opposite in direction. You are correct that F=ma plays a role though.
If m(cannon) = 100 and m(projectile) = 1, and the forces are equal, let's say F=1, then the acceleration a(cannon) will be 1/100 =
I say we transfer the program to the deparement of education on the grounds that it'll make for some good physics textbook examples.
I'm looking across my desk at my:
- 1 year old Dell 1.8 GHz P4
- 3 year old Apple Powerbook 400 Mhz G4
(similar RAM on both)
The G4 takes about twice as long to compile a large java project (5000 classes) as the Dell, which isn't bad for a 3 year old machine.
On the other hand, the Dell's DVD drive is broken, the latch for the battery keeps coming loose, and the machine gets hot and runs fan while it's sleeping.
The powerbook has intermittent problems with a clicking / scraping sound while watching DVDs. That's it. Battery life is still excelent, and the OS keeps getting faster with each point release.
Oddly enough, my mac (OS X) runs all of those forms/swing apps I've tried without a complaint even though it's obviously not running JInitiator and isn't certified to work. Same goes for their JDeveloper IDE. Given the apparent lack of JVM incompatability, I'd imagine they'll just do away with JInitiator.
One good reason you might want the database to have some say in the caching is to prevent cache pollution.
Oracle, for example, implements a split cache with hot and cold halves. Small reads get added to the head cache, while major reads (such as full table / index scans over a certain threshold) are added half way down the LRU stack in the "cold" half.
The idea here is that full scan type activity is less likely to provide cache hits on future queries. Also- full scan blocks come at a much lower cost per block than single, random access blocks, and therefore represent a lower value / storage ratio.
This is the kind of guess that's difficult for an operating system to make.
There's no special "compatibility" mode that I know of, except for the fact that it's a UNIX, and therefore very similar to Linux. Many apps can be ported with a simple compile (most of the GNU stuff just works). Oracle ported their entire database (although it's still considered beta) with minimal effort. In a commercial setting though, testing, documentation, and support will always cost real money.
I'm doing exactly what you're talking about using CVS.
.xyz binary files trying to do linefeed conversion, just because you forgot to tell it otherwise.
Everything I care about lives in huge CVS project, currently hosted on a laptop. In my case, this comes to 7GB, mostly in my photo gallery.
For servers (Solaris) at work, everything shares one NFS mounted home directory, which has a checkout of everything but my photos (due to quota limits). My personal work machine (Win2k, yuck) has its own complete checkout, with a few partial checkouts to put configuration files where different programs want them. My home machine (OS X) has a complete checkout.
This setup works for everything except bookmarks, where I find myself using IE on Windows, Netscape in Unix, and Safari on OS X. For bookmarks, I've moved to the "My Yahoo" service, which keeps everything online and accessable with one click from my home page.
By the way, I'd strongly recommend configuring CVS to default to binaries unless told otherwise for this kind of config. It's a nasty surprise to realize that CVS just mangled all of your
Store the Windows vulnerabilities on a Windows server, Linux vulnerabilities on a Linux server, etc.
That might take the edge off some companies' complaints about vulnerabilities leaking out before the clock is up.
Just chiming in with another data point:
My 1.5 year old TiBook 400 still gets 3 hours or so with full brightness. I generally only discharge down to 50% or so, but it's certainly been down to zero several dozen times.
This is much better than the Dell laptop I had in the past, which spent almost all of it's time plugged in but was only getting 30-45 minutes after a year of use. It also ran the fan intermittently while in standby, which never made much sense.
More than that, I'm always amazed at terrible coding pain non-CS people frequently subject themselves to by not using abstraction, source control, simple comments, or even descriptive names for variables and functions. This is in spite of the fact that many of these people do nothing but write code all day long.