I've always counted it as work. I have one simple rule. Am I as productive during the commute as at the office.
Sometimes the answer is yes and others it is no. If I feel I'm not overly productive I do not account for it.
Where this really kicks in hard is business travel. If I'm flying intercity then I usually end up doing a ton of work at the airport, on the shuttle or train to and from and some times on the plane it self.
This attitude instantly changed a few things. 1. I travel to the office at different times than the majority. Which means I'm more productive. Less distractions. 2. I travel cheaper. Off rush travel is often cheaper by air and public transport. ( Depends on city ) Company loves this and so do I. 3. Stress levels drop.
My "work week" is actually closer to 40 hours than most and I'm far more productive in those hours than my colleagues that are stressed way out.
Object Serialization was horrible from day one. It was the tool of lazy programmers for years.
Performance wise it was a disaster. People would pass objects between jvm instances with no regard to the size of the data blob they were sending. When in reality very little of an object is required to be sent in most use cases. Example: When you have a java cluster and you sync session objects between instances. ( Note this is just dumb anyway. There are far better patterns for this. ) But your low cost developer is stuff the session object with literally everything they can think of for data. So you see these monster 20Meg Objects fly around. The bigger you cluster gets the SLOWER it becomes as all the object passing is just simply taking over all the processing and memory.
Security. Massive issue here with object deserialization. Basically you can sneak any payload in you want here. You may have had value checks on the set methods but the vast majority simply trust a deserialized object. Trivial to change a string value into a sql injection string. And a million other holes just like this one.
Object serialization means that the two ends of the object share must now be in LOCK step with each wr.t. to class signatures. Which means to upgrade them you have to bring both end down at the same time. In the case of web infrastructure you are now forced to take that service completely offline in most scenarios. And if you are using something like RMI to a back end. That now has to be upgraded as well. This makes change operations incredibly painful for no reason whatsoever.
Personally I like the use of protobuf as a means of passing object data around. It literally solves all of the above problems with very little overhead. And it has the added benefit of being compatible with many other languages. It's also far tighter than xml and end thus takes less to transmit. Where as XML tends to cause a temp object storm on the receiver as it is deserialized. Which in turn causes more expensive garbage collection cycles.
One of the most significant differentiators is that when governments control the funing paths of industry they also control decision making in the industry.
SpaceX is independent and makes their own decisions. They just happen to have written a screw the feds contract that brings more money in.
NASA is a government run industrial institution. It's priorities are set by politicians. But in order to maintain funding other decisions are made to favor the politicians. For example where are the NASA jobs going to be located? The answer is a political one. Where are parts going to be developed, tested, assembled etc. All political answers.
The politically driven process is inherently more expensive. Simply because the most efficient and cheapest way to conduct business is usually not the chosen path.
With the Ariane 6 the proposal on the board is that Ariane plans to buy out the government stake in the company. Thus freeing it to directly compete on a level footing.
All credit to the government sponsored space programs over the decades. They created the seed tech and the science that is now being capitalized by the private industry.
I really think this is a strong release. The notable uplift of packages to directly tackle things like the spectre and meltdown defects.
However Ever since systemd-resolved was added to systemd in ubuntu 16.10 the resolver has been broken. Especially in local networks with dhcp and local dns.
It's basically un-usable as it stands. But a recent change in systemd-resolved makes it fairly trivial to get resolution working again.
With version 237 of systemd a stub-resolv.conf file was added. This is what the local system linked to./etc/resolv.conf ->../run/systemd/resolve/stub-resolv.conf This stub file tells everything to use the resolved process for dns resolution. It's broken so things don't work.
Simply change it to this and things work again./etc/resolv.conf ->../run/systemd/resolve/resolv.conf Because now they include a sensible resolv.conf as well as the broken version.
Big Brother Enterprise at the moment listens and gathers WAY to much data from those smart speakers.
- Some record and transmit everything they hear to their respective mothership. - Most are granted access to various social media and enterprise accounts.
The combination of access to all communications and data and passing of the data to the Enterprise mothership is a disaster in the making. This makes Enterprise a significant threat to our privacy and safety.
The other end of the issue is the immaturity of the devices themselves. These things are a hackers dream target. I don't need to outline the risks involved of a device that is always connected to our personal data.
The risk factors associated with these devices is way too high to even remotely contemplate inclusion into the home.
There is going to be a segment of the population that will purchase these devices. But those early adopters will quickly become the entire market. If there was such a pent up demand for the devices we would be seeing a much faster uptake of them. But the floundering Apple sales go to the point that the market demand is not at the much overhyped levels.
I'm basically saying is that the clock is two hours out year round. Yes the current system gives more sun in the summer months. But it basically means we come home in the dark in winter. Why not come home to a little sun and enjoy some outdoor time with the family, friends the garden what have you.
We should be moving the clock in winter and leaving it alone in summer. Even better move the whole clock an hour and apply reverse DST in winter thus the 2 hours.
Day Light Savings was put into place so that farmers children could help on the farm before heading off to school.
The farm day starts at sunrise.
All fine and dandy when the majority of the population was rural and farming was much much more manual labour than it is today.
Now however the balance has shifted to a urban population. Where tasks prior to work or school are minimal in comparison to that of the farmer decades before. Now with Daylight savings we are actually being robbed of 2 hours of potential daylight after work and school. Which in turn takes away time from extra curricular activities outside and takes away time from the family that doesn't involve screen time.
If any thing I think daylight savings should go the other way. Not just get rid of it.
If this tech is developed for Space travel or is an outcome of research into supporting us in space then it can most likely be ported into farming.
Who says this tech has to be used to feed humans directly. Why can't use it to improve the efficiency of animal husbandry. We can use farm wastes to more directly loop back into the production cycle. We could use it to improve the production of meat, textiles, milk, medical supplies etc.
Why not feed the outputs to say ants, and in return feed the ants to other stocks. Thus further diversifying the various protein chains making the process even more benificial.
Or we do more engineering and create organic polymers from the protein. Thus a replacement for many of the plastics in use today.
It does not have to directly feed us. As a matter of fact it can help reduce the costs associated with meat etc.
The key to the process is that it is a bio reactor that minimises infections and contamination. Thus making it more robust as an industrial system.
Linux desktop may very well become the only desktop in the future. Not because it won. It's because the other desktops died.
The only real use for a desktop now is for business use. Personal use of desktops is crashing. Mobile devices have effectively taken over personal use.
The browser has taken over as the OS on desktops. The applications are provided mostly by website interfaces. I have desktop machines that no longer have office suites installed, or graphical manipulation programs.
We will still see beefed up machines. But only for the purpose of running online application via the browser.
Personally I run Linux on basically every device attached to a monitor or TV as well as all my server gear. I have token windows and apple devices / vm's. But even a Linux fan boy like myself knows Linux desktop will never have it's big year. Simply because the desktop is dead.
30 years of experience means they actually know a lot more about the business than you might think.
Granted if they are unwilling to just leap on over the the current hipster computing model then maybe you have to figure something out.
These guys literally hold the companies legacy in their heads. Move them away from ops type tasks and onto data capture tasks. These people have seen every reason and method of storing legacy data over the years. Pick their brains and use it as input to migrate and obsolete legacy data stores and structures.
They more than likely also understand the business model over time better than most. Yeah they may be a horrible pain in the ass to talk to. But for a product manager they probably hold a mountain of info.
Oh I defend against known spying as well. This thread was all about IOT. So my previous post was targeting that conversation.
I block all traffic to analytics, ad, malware address. I do this multiple ways. The first is with my own internal DNS I change the resolution of these names to a dead end address. I update all lists daily from known lists on the net. I also pump those IP addresses into the firewall and block them there as well. This is probably an advanced topic for most people and most people would not be able to do this.
For a range of sites that I do use I run there routes through a proxy to strip certain content from the http/https streams. I do not recommend people do this. As it requires maintenance of cert's etc. This is very difficult to maintain. As this is an arms race against various web sites.
Note: My goal in all of this is to do most of the filtering and organisation of traffic at the network level. This enables me to add devices at will to the home network with a higher level of trust than most others.
In my home I implement two different networks. Each with it's own gate way. Now this requires more than your average level of IT skills in the home.
One network is for what I will call class one devices. These are devices that I specifically add to the it. These will be things like computers, tablets, gaming and phones. The second network and the default network is for every other device. Now this requires me actually promote devices the class one network. Typically be mac address.
Thus all those pesky iot devices end up in the default network. The default network is blocked from the internet.
Note a device that runs something like pfsense will do the job. There are lots of alternative setups.
Now. I can also tailor each device in each network to have slightly different network privileges than the each networks default. Example would be a security camera uploading data to my private cloud storage. But I also block all DNS resolution of add servers and malware end points etc in my class one network.
This is not something a regular I know how to turn on my laptop kinda person can do. This requires a reasonable amount of automated scripting, network monitoring and pro-active tuning as situations change. However it can all be done rather cheaply with couple hundred dollar pfsense box installed between the internet modem->pfsense->router(wifi).
So yeah I block everything. I only enable access when required and even then I can make it temporary. The more IOT crap that ends up in the house the more this setup is saving my backside.
( Note: I don't use pfsense I implemented all the services I need from pfsense myself in VM's. But it's basically the same thing. )
My experience in the field has led me to the conclusion that closed source is far more damaging to a companies bottom line.
1. Closed source licencing often results in changes to architectural designs to limit license exposure. This in turn often makes the final product weaker than it could be. For example if your buy of license X you can only scale to 10 nodes in production. If demand gets high enough you can not scale to meet it. 2. Closed source licenses that restrict functionality once license is exceeded. This can cripple your company and you don't even see it coming. If the product you have purchase has a version of cripple wear built in you could at times of demand all of a sudden lose functionality. Example, say the closed source stuff has a search function you need for your sales team. It's Xmas season and the sales are flying in. All of a sudden in the middle of the day the search function suddenly only returns the first match only. Chaos would ensue. 3. Support, Open source hands down has the best support out there. Opensource communities tend to be very active. This is a key requirement for any source you bring in house. I don't care if it's open or closed. I want to see the community behind it. Is it active. Are people passing examples around. Is there a friendly dialogue going on. Or is it just a stream of "This is broken". Or worse the last update was a year ago. Or even worse the community portal is managed by someone that deletes negative comments. Again Open source does this right. 4. No patches unless you pay a fee. The trap. You bought our stuff for cheap. After 3 years all of a sudden the support fee exceeds the original purchase price. Not all closed source does this. But it is a fairly common practice. They have your data / business by the short and curlies. If you don't pay you don't get support. Even worse you lose functionality if it is also crippleware. 5. Discontinued code. 5 years ago you built you business on this code you paid for. But now the company has decided to no longer develop it further. They aren't out of business and the claim to support it. But it's effectively dead. So that clause that says you get the source code if the business goes under still can't be invoked. Why because they are still there and they still say they support it. 6. Tool chains that you have no visibility of. Closed code also has closed tool chains. Which means you have no idea how it was built. Was it built on the interns laptop and hand rolled into a package? The intern that has a thing for surfing dangerous world of unicorn manga. His laptop that is so infected with malware that it takes 20 minutes to boot. You just don't know do you. Opensource the tool chain is typically part of the source. You can reproduce the build locally at any time or ever time.
Closed Source is extremely dangerous in my opinion. I only recommend closed source when the vendor is clearly the market segment leader with a strong community and a reputation for support. The number of closed products that fit this criteria is extremely small.
DRM blocks will now cause. - media blocks by region to become much more common. ( ala youtube video not available in your region ) - Sharing media links resulting in blocked content. - DRM collection of PII data becomes norm. Creating rich honey pots of data for hackers.
Oracle did not buy tech from Sun hardware or software.
Oracle bought customers and revenue streams. Or the potential for revenue streams in some cases.
Oracle knows that it can not keep support more and more software and hardware platforms. Oracle never mismanaged anything internally. They would have never intended to a lot of it going past the 2,5,10 year marks for various tech. They did mismanage or failed to anticipate market and client reaction. Case in point the Java non-sense where they tried to claw money, big money out of old sun clients that had no obligation to pay.
As for some of Sun's mistakes. The open source strategy was broken. Several reasons. They pushed the SPARC arch above all else. They kept tools required to build advanced product proprietary. Their community portals were horrible. Solaris had a broken patching/package system. ( God that was awful ) They simply did not adapt/adopt fast enough to be part of the Opensource world. yum and apt were light years ahead. They completely underestimated the power of creating a professional services unit that would actually be competent with opensource tools.
You are right though. The Oracle buy was the best deal they would ever see.
I foresee upcoming POWER releases as just another target for old LPAR consolidation until the apps that run on those systems can be retired or replatformed.
Ah good old LPAR. This was probably one of IBM's biggest successes. It allowed them to sell a lot of hardware. You basically had to massively over provision in order to use LPAR's. It also had a great security story in the sales pitch. They managed to replicate their success of statically allocated disk space on mainframes to statically allocating all other resources on the host as well.
LPAR's have for most application needs have been met by virtual machines. As the security/trust of VM's improved over the years so did their adoption stealing from LPAR's market. Now we are seeing the push of containers. Containers have a LONG way to go before they can be trusted. They also have to solve some nasty issues with networking and storage. A lot of issues actually.
AIX will limp on for at least 15 more years. The big reason is there are shops out there that simply are unable to migrate. Gov/Mil/Bank sectors will keep it alive for a long time. Hell they are currently keeping it alive.
The real issue is the death of the physical architecture. With x86 architecture becoming so powerful and low power and price points it's increasingly difficult justifying the purchase of outdated chips. Emulation now often outperforms actual physical chips.
Looking at recent massive threaded chips from intel and amd it's really hard to justify these old architectures. I can buy and i9 from intel for home server use that blows the doors off of anything in the office server room. And I can emulate basically any other chip on it. Same goes for amd's thread ripper.
Yes it would kill RHEL. And super fast. But again Oracle would be buying the customer base. I also suspect that Oracle would only certify RHEL on certain friendly hardware platforms. AKA ones that kick back to ORACLE for said certification.
CentOS would fork instantly and multiple other companies would try to model them selves after RedHat. Almost every RedHat project would fork as well. Git hub would go into meltdown for a few weeks I suspect.
The Debian variant's would all of a sudden see a lot more server room time. And good old FreeBSD would see a lot more loven. Please let FreeBSD see more love. BSD is the only hope to smash Cisco's lock on switching. It would also probably drive CoreOS well and truly into the mainstream. ( As it should be. )
Memory hogging java apps. Especially those that had to support lots of user sessions.
Generally you try to avoid heaps this big as garbage collection will kill the app. But sometimes you have to. Also with off heap memory methods now heaps can become enormous. We are talking Tbytes in size. I even know of a Pbyte implementation of offheap memory.
Obviously Oracle is clearing the decks to make room for something. The cash bleeders in the company are being sold off or shut down.
This leaves some big holes in Oracle. Things that helped prop it's DB business are now basically gone. Oracle has publicly gone all in with the cloud. Privately I think they have other plans.
So who is Oracle going to try and buy? Is it Red Hat? RHT 19.07B ORCL 209.399B
They could do it easily. They could buy control for 10B.
Or would they go after a global cloud provider? Like Digital Ocean?
Slashdot Mirror
I've always counted it as work. I have one simple rule. Am I as productive during the commute as at the office.
Sometimes the answer is yes and others it is no. If I feel I'm not overly productive I do not account for it.
Where this really kicks in hard is business travel. If I'm flying intercity then I usually end up doing a ton of work at the airport, on the shuttle or train to and from and some times on the plane it self.
This attitude instantly changed a few things.
1. I travel to the office at different times than the majority. Which means I'm more productive. Less distractions.
2. I travel cheaper. Off rush travel is often cheaper by air and public transport. ( Depends on city ) Company loves this and so do I.
3. Stress levels drop.
My "work week" is actually closer to 40 hours than most and I'm far more productive in those hours than my colleagues that are stressed way out.
Seriously.
This is all wrong. Young people will not understand what they are giving away here.
Object Serialization was horrible from day one. It was the tool of lazy programmers for years.
Performance wise it was a disaster. People would pass objects between jvm instances with no regard to the size of the data blob they were sending. When in reality very little of an object is required to be sent in most use cases. Example: When you have a java cluster and you sync session objects between instances. ( Note this is just dumb anyway. There are far better patterns for this. ) But your low cost developer is stuff the session object with literally everything they can think of for data. So you see these monster 20Meg Objects fly around. The bigger you cluster gets the SLOWER it becomes as all the object passing is just simply taking over all the processing and memory.
Security. Massive issue here with object deserialization. Basically you can sneak any payload in you want here. You may have had value checks on the set methods but the vast majority simply trust a deserialized object. Trivial to change a string value into a sql injection string. And a million other holes just like this one.
Object serialization means that the two ends of the object share must now be in LOCK step with each wr.t. to class signatures. Which means to upgrade them you have to bring both end down at the same time. In the case of web infrastructure you are now forced to take that service completely offline in most scenarios. And if you are using something like RMI to a back end. That now has to be upgraded as well. This makes change operations incredibly painful for no reason whatsoever.
Personally I like the use of protobuf as a means of passing object data around. It literally solves all of the above problems with very little overhead. And it has the added benefit of being compatible with many other languages. It's also far tighter than xml and end thus takes less to transmit. Where as XML tends to cause a temp object storm on the receiver as it is deserialized. Which in turn causes more expensive garbage collection cycles.
One of the most significant differentiators is that when governments control the funing paths of industry they also control decision making in the industry.
SpaceX is independent and makes their own decisions. They just happen to have written a screw the feds contract that brings more money in.
NASA is a government run industrial institution. It's priorities are set by politicians. But in order to maintain funding other decisions are made to favor the politicians. For example where are the NASA jobs going to be located? The answer is a political one. Where are parts going to be developed, tested, assembled etc. All political answers.
The politically driven process is inherently more expensive. Simply because the most efficient and cheapest way to conduct business is usually not the chosen path.
With the Ariane 6 the proposal on the board is that Ariane plans to buy out the government stake in the company. Thus freeing it to directly compete on a level footing.
All credit to the government sponsored space programs over the decades. They created the seed tech and the science that is now being capitalized by the private industry.
I got the email.
I was impressed that it was handled quickly.
I'm even more confident because I actually use a proper password manager making sure I have unique passwd's for everything.
I really think this is a strong release. The notable uplift of packages to directly tackle things like the spectre and meltdown defects.
However
Ever since systemd-resolved was added to systemd in ubuntu 16.10 the resolver has been broken. Especially in local networks with dhcp and local dns.
It's basically un-usable as it stands. But a recent change in systemd-resolved makes it fairly trivial to get resolution working again.
With version 237 of systemd a stub-resolv.conf file was added. This is what the local system linked to. /etc/resolv.conf -> ../run/systemd/resolve/stub-resolv.conf
This stub file tells everything to use the resolved process for dns resolution. It's broken so things don't work.
Simply change it to this and things work again. /etc/resolv.conf -> ../run/systemd/resolve/resolv.conf
Because now they include a sensible resolv.conf as well as the broken version.
Big Brother Enterprise at the moment listens and gathers WAY to much data from those smart speakers.
- Some record and transmit everything they hear to their respective mothership.
- Most are granted access to various social media and enterprise accounts.
The combination of access to all communications and data and passing of the data to the Enterprise mothership is a disaster in the making. This makes Enterprise a significant threat to our privacy and safety.
The other end of the issue is the immaturity of the devices themselves. These things are a hackers dream target. I don't need to outline the risks involved of a device that is always connected to our personal data.
The risk factors associated with these devices is way too high to even remotely contemplate inclusion into the home.
There is going to be a segment of the population that will purchase these devices. But those early adopters will quickly become the entire market. If there was such a pent up demand for the devices we would be seeing a much faster uptake of them. But the floundering Apple sales go to the point that the market demand is not at the much overhyped levels.
It happened during Bush's presidency.
Yep
I should have elaborated.
I'm basically saying is that the clock is two hours out year round. Yes the current system gives more sun in the summer months. But it basically means we come home in the dark in winter. Why not come home to a little sun and enjoy some outdoor time with the family, friends the garden what have you.
We should be moving the clock in winter and leaving it alone in summer. Even better move the whole clock an hour and apply reverse DST in winter thus the 2 hours.
Day Light Savings was put into place so that farmers children could help on the farm before heading off to school.
The farm day starts at sunrise.
All fine and dandy when the majority of the population was rural and farming was much much more manual labour than it is today.
Now however the balance has shifted to a urban population. Where tasks prior to work or school are minimal in comparison to that of the farmer decades before. Now with Daylight savings we are actually being robbed of 2 hours of potential daylight after work and school. Which in turn takes away time from extra curricular activities outside and takes away time from the family that doesn't involve screen time.
If any thing I think daylight savings should go the other way. Not just get rid of it.
If this tech is developed for Space travel or is an outcome of research into supporting us in space then it can most likely be ported into farming.
Who says this tech has to be used to feed humans directly. Why can't use it to improve the efficiency of animal husbandry. We can use farm wastes to more directly loop back into the production cycle. We could use it to improve the production of meat, textiles, milk, medical supplies etc.
Why not feed the outputs to say ants, and in return feed the ants to other stocks. Thus further diversifying the various protein chains making the process even more benificial.
Or we do more engineering and create organic polymers from the protein. Thus a replacement for many of the plastics in use today.
It does not have to directly feed us. As a matter of fact it can help reduce the costs associated with meat etc.
The key to the process is that it is a bio reactor that minimises infections and contamination. Thus making it more robust as an industrial system.
Linux desktop may very well become the only desktop in the future. Not because it won. It's because the other desktops died.
The only real use for a desktop now is for business use. Personal use of desktops is crashing. Mobile devices have effectively taken over personal use.
The browser has taken over as the OS on desktops. The applications are provided mostly by website interfaces. I have desktop machines that no longer have office suites installed, or graphical manipulation programs.
We will still see beefed up machines. But only for the purpose of running online application via the browser.
Personally I run Linux on basically every device attached to a monitor or TV as well as all my server gear. I have token windows and apple devices / vm's. But even a Linux fan boy like myself knows Linux desktop will never have it's big year. Simply because the desktop is dead.
No they can't. A nuke plant takes days to ramp up electricity production. A nuke plant is not an instant on by any means.
A Nuke plant is a base load provider not a peak provider. It is near impossible to design a fast ramp up nuke plant. As in with in 30 minutes.
Gas and Hydro are near instantaneous power providers. Hydro being the fastest to provide power.
Exactly ask them.
30 years of experience means they actually know a lot more about the business than you might think.
Granted if they are unwilling to just leap on over the the current hipster computing model then maybe you have to figure something out.
These guys literally hold the companies legacy in their heads. Move them away from ops type tasks and onto data capture tasks. These people have seen every reason and method of storing legacy data over the years. Pick their brains and use it as input to migrate and obsolete legacy data stores and structures.
They more than likely also understand the business model over time better than most. Yeah they may be a horrible pain in the ass to talk to. But for a product manager they probably hold a mountain of info.
Oh I defend against known spying as well. This thread was all about IOT. So my previous post was targeting that conversation.
I block all traffic to analytics, ad, malware address. I do this multiple ways. The first is with my own internal DNS I change the resolution of these names to a dead end address. I update all lists daily from known lists on the net. I also pump those IP addresses into the firewall and block them there as well. This is probably an advanced topic for most people and most people would not be able to do this.
For a range of sites that I do use I run there routes through a proxy to strip certain content from the http/https streams. I do not recommend people do this. As it requires maintenance of cert's etc. This is very difficult to maintain. As this is an arms race against various web sites.
Note: My goal in all of this is to do most of the filtering and organisation of traffic at the network level. This enables me to add devices at will to the home network with a higher level of trust than most others.
In my home I implement two different networks. Each with it's own gate way. Now this requires more than your average level of IT skills in the home.
One network is for what I will call class one devices. These are devices that I specifically add to the it. These will be things like computers, tablets, gaming and phones. The second network and the default network is for every other device. Now this requires me actually promote devices the class one network. Typically be mac address.
Thus all those pesky iot devices end up in the default network. The default network is blocked from the internet.
Note a device that runs something like pfsense will do the job. There are lots of alternative setups.
Now. I can also tailor each device in each network to have slightly different network privileges than the each networks default. Example would be a security camera uploading data to my private cloud storage. But I also block all DNS resolution of add servers and malware end points etc in my class one network.
This is not something a regular I know how to turn on my laptop kinda person can do. This requires a reasonable amount of automated scripting, network monitoring and pro-active tuning as situations change. However it can all be done rather cheaply with couple hundred dollar pfsense box installed between the internet modem->pfsense->router(wifi).
So yeah I block everything. I only enable access when required and even then I can make it temporary. The more IOT crap that ends up in the house the more this setup is saving my backside.
( Note: I don't use pfsense I implemented all the services I need from pfsense myself in VM's. But it's basically the same thing. )
My experience in the field has led me to the conclusion that closed source is far more damaging to a companies bottom line.
1. Closed source licencing often results in changes to architectural designs to limit license exposure. This in turn often makes the final product weaker than it could be. For example if your buy of license X you can only scale to 10 nodes in production. If demand gets high enough you can not scale to meet it.
2. Closed source licenses that restrict functionality once license is exceeded. This can cripple your company and you don't even see it coming. If the product you have purchase has a version of cripple wear built in you could at times of demand all of a sudden lose functionality. Example, say the closed source stuff has a search function you need for your sales team. It's Xmas season and the sales are flying in. All of a sudden in the middle of the day the search function suddenly only returns the first match only. Chaos would ensue.
3. Support, Open source hands down has the best support out there. Opensource communities tend to be very active. This is a key requirement for any source you bring in house. I don't care if it's open or closed. I want to see the community behind it. Is it active. Are people passing examples around. Is there a friendly dialogue going on. Or is it just a stream of "This is broken". Or worse the last update was a year ago. Or even worse the community portal is managed by someone that deletes negative comments. Again Open source does this right.
4. No patches unless you pay a fee. The trap. You bought our stuff for cheap. After 3 years all of a sudden the support fee exceeds the original purchase price. Not all closed source does this. But it is a fairly common practice. They have your data / business by the short and curlies. If you don't pay you don't get support. Even worse you lose functionality if it is also crippleware.
5. Discontinued code. 5 years ago you built you business on this code you paid for. But now the company has decided to no longer develop it further. They aren't out of business and the claim to support it. But it's effectively dead. So that clause that says you get the source code if the business goes under still can't be invoked. Why because they are still there and they still say they support it.
6. Tool chains that you have no visibility of. Closed code also has closed tool chains. Which means you have no idea how it was built. Was it built on the interns laptop and hand rolled into a package? The intern that has a thing for surfing dangerous world of unicorn manga. His laptop that is so infected with malware that it takes 20 minutes to boot. You just don't know do you. Opensource the tool chain is typically part of the source. You can reproduce the build locally at any time or ever time.
Closed Source is extremely dangerous in my opinion. I only recommend closed source when the vendor is clearly the market segment leader with a strong community and a reputation for support. The number of closed products that fit this criteria is extremely small.
( Sorry spelling is horrible. )
XP is a security nightmare. It was a great OS in it's time but that was well almost 18 years ago.
I just wish more websites would simply block old outof support OS's. It's not overly hard to block.
Firefox is only enabling horrible security practice by doing this so late.
DRM blocks will now cause.
- media blocks by region to become much more common. ( ala youtube video not available in your region )
- Sharing media links resulting in blocked content.
- DRM collection of PII data becomes norm. Creating rich honey pots of data for hackers.
Oracle did not buy tech from Sun hardware or software.
Oracle bought customers and revenue streams. Or the potential for revenue streams in some cases.
Oracle knows that it can not keep support more and more software and hardware platforms. Oracle never mismanaged anything internally. They would have never intended to a lot of it going past the 2,5,10 year marks for various tech. They did mismanage or failed to anticipate market and client reaction. Case in point the Java non-sense where they tried to claw money, big money out of old sun clients that had no obligation to pay.
As for some of Sun's mistakes.
The open source strategy was broken. Several reasons. They pushed the SPARC arch above all else. They kept tools required to build advanced product proprietary. Their community portals were horrible. Solaris had a broken patching/package system. ( God that was awful ) They simply did not adapt/adopt fast enough to be part of the Opensource world. yum and apt were light years ahead. They completely underestimated the power of creating a professional services unit that would actually be competent with opensource tools.
You are right though. The Oracle buy was the best deal they would ever see.
I foresee upcoming POWER releases as just another target for old LPAR consolidation until the apps that run on those systems can be retired or replatformed.
Ah good old LPAR. This was probably one of IBM's biggest successes. It allowed them to sell a lot of hardware. You basically had to massively over provision in order to use LPAR's. It also had a great security story in the sales pitch. They managed to replicate their success of statically allocated disk space on mainframes to statically allocating all other resources on the host as well.
LPAR's have for most application needs have been met by virtual machines. As the security/trust of VM's improved over the years so did their adoption stealing from LPAR's market. Now we are seeing the push of containers. Containers have a LONG way to go before they can be trusted. They also have to solve some nasty issues with networking and storage. A lot of issues actually.
AIX will limp on for at least 15 more years. The big reason is there are shops out there that simply are unable to migrate. Gov/Mil/Bank sectors will keep it alive for a long time. Hell they are currently keeping it alive.
The real issue is the death of the physical architecture. With x86 architecture becoming so powerful and low power and price points it's increasingly difficult justifying the purchase of outdated chips. Emulation now often outperforms actual physical chips.
Looking at recent massive threaded chips from intel and amd it's really hard to justify these old architectures. I can buy and i9 from intel for home server use that blows the doors off of anything in the office server room. And I can emulate basically any other chip on it. Same goes for amd's thread ripper.
Yes it would kill RHEL. And super fast. But again Oracle would be buying the customer base. I also suspect that Oracle would only certify RHEL on certain friendly hardware platforms. AKA ones that kick back to ORACLE for said certification.
CentOS would fork instantly and multiple other companies would try to model them selves after RedHat. Almost every RedHat project would fork as well. Git hub would go into meltdown for a few weeks I suspect.
The Debian variant's would all of a sudden see a lot more server room time. And good old FreeBSD would see a lot more loven. Please let FreeBSD see more love. BSD is the only hope to smash Cisco's lock on switching. It would also probably drive CoreOS well and truly into the mainstream. ( As it should be. )
That's exactly the reason.
Memory hogging java apps. Especially those that had to support lots of user sessions.
Generally you try to avoid heaps this big as garbage collection will kill the app. But sometimes you have to. Also with off heap memory methods now heaps can become enormous. We are talking Tbytes in size. I even know of a Pbyte implementation of offheap memory.
Obviously Oracle is clearing the decks to make room for something. The cash bleeders in the company are being sold off or shut down.
This leaves some big holes in Oracle. Things that helped prop it's DB business are now basically gone. Oracle has publicly gone all in with the cloud. Privately I think they have other plans.
So who is Oracle going to try and buy? Is it Red Hat?
RHT 19.07B
ORCL 209.399B
They could do it easily. They could buy control for 10B.
Or would they go after a global cloud provider? Like Digital Ocean?