Jump to content

Falconevo

Member
  • Posts

    906
  • Joined

  • Last visited

Reputation Activity

  1. Agree
    Falconevo got a reaction from LAwLz in Backblaze: SSDs might be as unreliable as disk drives   
    SSD's are consumables, people really need to digest and understand this.  Eat their rated write endurance and it will be a door stop in no time.

    What bugs me is the article from BackBlaze is missing vital information, because i am assuming that they buy consumer grade SSD's and mechanical disks then throw them in to the fray expecting enterprise levels of endurance... 😕 not sure what else they would expect besides early failure rates.

    The article is incredibly misleading, an SSD will die within weeks if you obliterate it's potato flash write endurance in that time.

    What the article is missing
    What SSD models they were using and why What the budget per server was for the SSDs vs Mechanical drives What was the TBW at the point of failure (was it inside or outside manufacturer rated spec) Were SSDs behind a RAID/HBA controller with/without trim support and was the OS in use providing said support Are they utilising or reading the additional manufacturer specific S.M.A.R.T data provided As someone who has put over 50k SSD's in to environments from consumer to DC grade devices I find the whole thing misleading. I have seen numerous SSD's fail for all sorts of reasons but the biggest one by far is exceeding the manufacturer rated endurance, followed by Intel's 'DISABLE LOGICAL STATE' which still gives me nightmares to this day.
     
  2. Agree
    Falconevo got a reaction from leadeater in Backblaze: SSDs might be as unreliable as disk drives   
    SSD's are consumables, people really need to digest and understand this.  Eat their rated write endurance and it will be a door stop in no time.

    What bugs me is the article from BackBlaze is missing vital information, because i am assuming that they buy consumer grade SSD's and mechanical disks then throw them in to the fray expecting enterprise levels of endurance... 😕 not sure what else they would expect besides early failure rates.

    The article is incredibly misleading, an SSD will die within weeks if you obliterate it's potato flash write endurance in that time.

    What the article is missing
    What SSD models they were using and why What the budget per server was for the SSDs vs Mechanical drives What was the TBW at the point of failure (was it inside or outside manufacturer rated spec) Were SSDs behind a RAID/HBA controller with/without trim support and was the OS in use providing said support Are they utilising or reading the additional manufacturer specific S.M.A.R.T data provided As someone who has put over 50k SSD's in to environments from consumer to DC grade devices I find the whole thing misleading. I have seen numerous SSD's fail for all sorts of reasons but the biggest one by far is exceeding the manufacturer rated endurance, followed by Intel's 'DISABLE LOGICAL STATE' which still gives me nightmares to this day.
     
  3. Agree
    Falconevo got a reaction from Abdullah Bhutta in SSD "corrupted", but still works (kind of)   
    I would suggest checking the write endurance left on the drive, people seem to forget that SSD's are a consumable.
  4. Agree
    Falconevo reacted to leadeater in SSD Revisions of same product show dissimilar and lacking performance   
    Such is why I only buy Samsung SSDs. Pitty as I thought the SX8200 Pro was actually extremely good and very very good value.
  5. Agree
    Falconevo reacted to leadeater in AMD EPYC on Mini-ITX - ASRock did it again   
    TR and Ryzen both do actually support ECC memory, what they do not support is Registered and Load Reduced ECC (RDIMM & LRDIMM). So you can use ECC UDIMM just fine, but these do not support larger capacity DIMM sizes.
  6. Agree
    Falconevo reacted to StDragon in Youtube making ads non-optional, will not pay non-partners   
    "The beating will continue until morale improves" -Youtube
  7. Like
    Falconevo got a reaction from Electronics Wizardy in Looking for Advise Turn GamingPC into unRAID NAS & VM for Gaming   
    I'm with @Electronics Wizardy here, this is going to be a whole load of pain for very little gain unless you are solely doing this to expand your knowledge.
  8. Agree
    Falconevo got a reaction from ne10g in Best router for 10G internet?   
    Yea it can be virtualised but you are not going to get 10Gbit throughput from a virtualised network interface.  You will have to use PCIe pass-through for the network interface(s) to reach that and place the interface in the dataplane
  9. Informative
    Falconevo got a reaction from ne10g in Best router for 10G internet?   
    Build a custom TNSR installation, its free now for home/non-commercial use.  This is what the pfSense guys have been working on along side the pfSense project but uses a different way of handling traffic using VPP (Vector Packet Processing) and DPDK (Data Plane Development Kit) to minimise CPU cycles.  These are developments on Linux to substantially improve per core packet processing capability.  From around 1Gbit/s per core to around 10Gbit/s per core.
     
    https://www.tnsr.com/
    https://www.tnsr.com/tnsr-vs-pfsense-software
     
    Can do 10G with ease, I would probably use Intel 10G, Mellanox Connect-X or SolarFlare cards in the installation.  Can pick them up pretty cheap, I'm using a SolarFlare SFN5121 in my colo/test chassis running TNSR.

    Has no GUI, but does have a REST API and console access, can get it setup for normal internet usage pretty easily and may teach you something new.
  10. Like
    Falconevo got a reaction from danwat1234 in How bad is this $5000 PC from 10 years ago?   
    Anyone able to clear something up?  I sent Linus this exact board, 2x QX9775 CPU's and some FB-Dimms from the UK @LinusTech



    Is this my board being used here as the box looks like the one I sent across or is this @Overl0rd's equipment?  Trying to figure out what has gone on here?  Top job for Overlord if he did send another board, cpu's etc but I don't get where the board and equipment I sent has gone....?

    Anyone from Linus Media Group able to clarify whats going on here?  Just wanna know if the stuff I sent was working etc
  11. Like
    Falconevo got a reaction from Shreck Ye in Sharing host-GPU with VM (qemu+kvm)   
    You cannot share a single GPU to multiple VM's without driver support from Nvidia Grid or AMD mxGPU.  You would need a card which has support for these drivers which are enterprise/DC grade equipment.  They are very expensive and it is doubtful you can get one for cheap.

    If you are using consumer grade equipment, they won't function with Grid or mxGPU drivers.
  12. Like
    Falconevo got a reaction from Lurick in anti-dosing protectiondatacenter   
    You should really speak to network specialists about this and the ISPs you plan to be peering with to see what they do/don't support.

    I can suggest some additional reading for you;
    FastNetMon for detection based on custom parameters and upstream black holes to reduce network infrastructure impact
    BGPFlowSpec for more advanced upstream black hole rules to allow known good traffic types etc
     
    I would also suggest you to avoid trying to mitigate/filter attacks until you have 200Gbit/s+ of internet bandwidth capacity at a data centre.  Considering you are asking this question, it is doubtful you have capacity to handle even a small attack.  Write it in to your terms & conditions regarding what you would do in the event of an attack.
     
  13. Agree
    Falconevo got a reaction from LAwLz in Intel passing off bogus Xeon Server 'metrics'   
    Not having a Windows license makes absolutely no different to features or performance on Windows Server OS
     
    Plus its only on that temporarily until it goes it to production, but its under testing currently anyway
  14. Like
    Falconevo got a reaction from Fnige in Intel passing off bogus Xeon Server 'metrics'   
    No complaints here from AMD's current offerings, 2.96Ghz on ALL cores for 2 fully loaded 7452 Rome CPUs .


  15. Agree
    Falconevo got a reaction from leadeater in Intel passing off bogus Xeon Server 'metrics'   
    Thats only part of the testing suite, thats the MSSQL testing which can be done on Linux but I certainly wouldn't want to use Linux for MSSQL at the moment.
  16. Like
    Falconevo got a reaction from SpaceGhostC2C in Intel passing off bogus Xeon Server 'metrics'   
    Thats only part of the testing suite, thats the MSSQL testing which can be done on Linux but I certainly wouldn't want to use Linux for MSSQL at the moment.
  17. Like
    Falconevo got a reaction from cj09beira in Intel passing off bogus Xeon Server 'metrics'   
    No complaints here from AMD's current offerings, 2.96Ghz on ALL cores for 2 fully loaded 7452 Rome CPUs .


  18. Like
    Falconevo got a reaction from leadeater in Intel passing off bogus Xeon Server 'metrics'   
    No complaints here from AMD's current offerings, 2.96Ghz on ALL cores for 2 fully loaded 7452 Rome CPUs .


  19. Informative
    Falconevo got a reaction from Fyfey96 in Pf Sense Custom Router   
    It wont see the traffic if the switch port being used has not been set to trunk and the vlans on the trunk port are allowed in the port configuration.   This carries for all devices attached to the switch that require multiple (tagged) vlans per port.
  20. Funny
    Falconevo reacted to leadeater in Intel gets EPYC-ally annihilated - Benchmarks & Testing (2x 64 "ROME" cores)   
    I'd rather watch 1 gamer 128 games. Can you handle playing multiple games at once? Lets find out.
  21. Like
    Falconevo got a reaction from Overl0rd in How bad is this $5000 PC from 10 years ago?   
    If memory serves it was able to run 1.38 volts to get 4.4 stable (silicon lottery?) I did have an excel document full of configuration settings used for each step of the way but I genuinely can't seem to locate it on any old hdd  

    I had a lot of problems with memory compatibility, I ended up with some Kingston 800Mhz FB-DIMMS KTA-MP800K2 and had to lax off the timings to keep things stable due to the FSB bump.  I also had a corsair active memory cooler on the top as the memory modules run HOT.
     
    The whole system ran hot and the power draw was colossal but it was a really fun system to build, I've always been an enthusiast for dual socket builds as I have always worked primarily with server hardware with dual and quad socket mainboards.   Speaking of such, I do have a pair of X5492's which OC'd really well, I think they surpassed the QX9775's with similar voltage, I still have those knocking around in the basement if they are of any interest to you?  Got a load of old hardware like that just sat in the basement doing nothing.
  22. Like
    Falconevo got a reaction from SPARTAN VI in How bad is this $5000 PC from 10 years ago?   
    Anyone able to clear something up?  I sent Linus this exact board, 2x QX9775 CPU's and some FB-Dimms from the UK @LinusTech



    Is this my board being used here as the box looks like the one I sent across or is this @Overl0rd's equipment?  Trying to figure out what has gone on here?  Top job for Overlord if he did send another board, cpu's etc but I don't get where the board and equipment I sent has gone....?

    Anyone from Linus Media Group able to clarify whats going on here?  Just wanna know if the stuff I sent was working etc
  23. Funny
    Falconevo reacted to porina in Quake II RTX is out now for free   
    If you mean performance will suck I'd agree, but I wouldn't call it a nerf which to me is an intentional crippling of performance.
     
     
    Also:
     
  24. Informative
    Falconevo got a reaction from Dedayog in Most reliable SSD brand?   
    Yea, every disk, HBA, array and storage appliances are heavily monitored, that's not to say failure's don't occur but we do have early warning systems for certain scenarios.  Most of the systems are built in house as we require integration to other systems and in house API's.
     
    We also audit for known issues with disk firmware versions, so if an issue crops up we can quickly audit systems to pull an inventory of potentially affected systems and take steps to arrange the necessary work to make sure known issues don't cause unwanted data loss.  Most work in the enterprise/DC sphere can be performed without down time but it all depends on the original budget of the solution and how much penny pinching occurred, the customer should never be out of pocket for problems that aren't of their own making (in my humble opinion)

    To give you a publicly known example, the Intel DC *4600 series suffered from a catastrophic issue on early firmware releases where the drive would enter 'disable logical state' under a number of assert or bad_context conditions .  The drive(s) would essentially be bricked and this could happen months after release in to production with no warning.. problems that are not always discovered in testing phases.  This caused no end of drama as the units are very popular and their successors are to this day so a lot of work was required to make sure we didn't have disks nuking themselves in storage appliances.
     
    Only Intel can recover the drive and I work in an industry that has different RMA agreements so the drive essentially have to be replaced and original destroyed rather than recovered due to potentially sensitive data.  Hate to see expensive SSD's hit the shredder but there's no alternative and it is heartbreaking for me to see ?... I don't mean to single out Intel here, similar incidents have happened with other manufacturers.  Not every product released is perfect straight out of the production line but good communication with the vendors and having technical contacts on their end makes a whole world of difference to getting problems sorted quickly.

    I've also built a number of systems for SSD stock ingest for firmware updating SSDs so we can make sure that drives being earmarked for use have the latest (if no known issues) and/or the most suitable firmware before going in to a production environments even if they have been in stock for a period of time.
  25. Like
    Falconevo got a reaction from dalekphalm in Plex port forwarding.   
    Change 'plex 2' to be 

    ID 1
    Service Type - plex 2
    External Port - 32401
    Internal IP - 192.168.1.168
    Internal Port - 32400
    Protocol - TCP (UDP is not required for Plex)
     
    Then remove ID 5 from the list, it is not required.
×