Kitz Forum
Computers & Hardware => Networking => Topic started by: Weaver on January 14, 2022, 07:26:25 AM
-
Say you have 10GBps or 40Gbps FTTP. What kind of router and switch tickles your fancy?
-
10G a Mikrotik CCR2004-1G-12S+2XS is fine.
40G a PC with an accelerator card, and a switch with a bunch of 40G ports.
Shifting that many packets per second is challenging.
-
Thank you. This is why a mere 1500 byte L3 PDU size is not good. A much much larger size would take the load off. One of the virtues of Token Ring.
-
When I'm home I'll have a look at the split between packet sizes on my broadband. I think most of it comes in below 1500.
It is overdue an increase for sure. The 1500 was based around far skinnier, more lossy networks.
-
Wouldn't larger packets cause a latency penalty? As some servers are thousands of miles away, more latency is not great.
Its also assuming no losses on the wider network which there will always be. One dropped packet of 1500 is far less problematic than say 9000.
Even on 10Gbit LAN the benefits of larger packets seem minimal in my experience.
-
I can’t understand how PDU size affects latency. That doesn’t seem at all right?
One other point is that of efficiency. With IPv6+TCP headers totalling 60 bytes or 72 with timestamps, that’s 4% or 4.8% of a 1500 byte PDU, but increasing the size of the PDU massively decreases this wastage, down to a very small fraction of a big packet.
-
I was thinking along the lines that if you are sending bigger packets then any new data has longer to wait for that packet to finish being sent, before another packet can be. Apparently this is indeed a thing (https://blogs.keysight.com/blogs/tech/nwvs.entry.html/2020/05/01/packet_sizes_formar-7oXg.html).
Also like I said before, if you lose a packet you have potentially a LOT more data to re-transmit, though obviously low latency protocols would just continue to use smaller packet sizes to avoid this. But if those smaller packets are stuck behind a queue of larger packets, you're so out of luck unless the router prioritises smaller packets, and then you have the issue that you've negated the speed benefit as larger packets have to wait for smaller packets. Then what's to stop me using smaller packets to get a leg-up on the delivery time?
I get this somewhat applies today as smaller packets are often prioritised, but were talking much smaller packets where if you chose to use those deliberately you would be slowing things down anyway, whereas using 1500 to get priority over say 9000, would likely be worth it.
But the big problem why it probably wont ever happen is while it reduces the CPU load on the client and server, those are the cheapest bits of hardware to upgrade. As there will always be legacy hardware on the Internet, routers having to fragment those packets would likely negate or exceed any benefit from those frames being more efficient when not fragmented. As you know, core routers are darned expensive so anything that might increase their CPU load is a bad idea, its computationally harder to fragment a larger packet than it is to just deliver several 1500 packets equalling the same total size.
The general wisdom these days seems to be to not use jumbo frames, even on a LAN, with the possible exception of over dedicated iSCSI links. Its more trouble than its worth. Now try multiplying that trouble by the entire Internet.
If you never ever were going to lose any packets and your interconnects never have contention, sure it would be more efficient, but that's not how the Internet works. We only need to look at ping graphs to see how often small ICMP packets fail to arrive in a timely manner, even on a connection that is idle.
-
Wouldn't larger packets cause a latency penalty?
Why? It's only increasing the maximum packet size allowed that's being suggested. Smaller packets (which is what is generally the case with protocols where latency is a concern) will still take exactly the same time.
-
But the big problem why it probably wont ever happen is while it reduces the CPU load on the client and server, those are the cheapest bits of hardware to upgrade. As there will always be legacy hardware on the Internet, routers having to fragment those packets would likely negate or exceed any benefit from those frames being more efficient when not fragmented. As you know, core routers are darned expensive so anything that might increase their CPU load is a bad idea, its computationally harder to fragment a larger packet than it is to just deliver several 1500 packets equalling the same total size.
Path MTU discovery with fallback if in doubt takes care of this. On the matter of queuing cable networks and their mechanisms to handle media acquisition are a good place to look. With how large upstream pipes are getting the transceiver time taken up by 9000 bytes relative to 1512 shrinks.
It's a situational thing, though. The capability for some apps to be able to use larger sizes would be good. Migrating everything over not necessarily. At really high throughputs ASICs are doing the work and these continue to be parallelised and get faster.
No doubt a bunch of the data we send goes into a router, is sent to a router next to it for more work as that's where the fabric line cards are, then returned to the original one for transmission out.
-
Obviously its hard to predict, I mean I've already read a document detailing on why all ISPs should move to DHCP instead of PPPoE, yet PPPoE rattles on. Not because its efficient, but because its more readily compatible with their legacy accounting systems.
Yet that should be much easier to deal with as it has zero impact on anything outside the ISP (except their customers of course), whereas jumbo frames are not part of the IEEE ethernet standard so you're kinda playing with fire.
Just to refresh my memory I just did a few tests right now between my NAS and Desktop. With a 9000 MTU its faster pushing data to my NAS with peaks up to 9.8Gbit compared to 9.2Gbit at 1500 which sounds great. However pulling data from it is actually slower at 6.1 to 7.3 vs 6.5 to 7.9. That was with iperf3 but NFS seems to concur where even though it doesn't get near that rate due to bottlenecking on SATA, its slightly faster with a normal MTU.
-
Talking about the PPP and DHCP thing, does that mean DHCPv4 (ie not DHCPv6). What do we do about IPv6-only users? And what do we do for users such as myself who use PPP LCP ping routers instead? That would assume that all applications are IP, or IPv4/6, whereas PPP caters for a variety of L3 protocols, not just IPv4 and IPv6. I don’t know why some network operators, such as BT, don’t like PPP. PPP is encapsulated in L2TP in the core network anyway, and L2TP is just another random packet payload (SDU) to be shifted in IP, so I can’t see how they even know or care. In my own case, it’s PPP to the router and then PPP all the way from router to ISP, PPP being carried in L2TP between BT’s BRAS and the ISP.
-
You're over thinking this.
DHCP in this case is just IPoE.
The PPP terminates at the BRAS. What can you do with PPP that you can't do without it?
PPPoE is just IP packets encapsulated in PPP, encapsulated in Ethernet.
IPoE is just IP packets encapsulated in Ethernet.
PPP adds overhead. It just gives your router more work to do. On some hardware PPP increases CPU usage quite dramatically.
The reason PPP is used is it makes it easier to wholesale.
It has no place on ultrafast broadband.
-
J0hn, I hear you. To answer your question, PPP LCP is used in AA’s network to provide their constant quality monitoring, but this is not much of an argument as their ping boxes can do the same thing with IP/ICMP and do so when used by the thinkbroadband and f8lure.mouselike.org monitoring services. I think we are at cross purposes as I was talking about core routers in carriers whereas I think you’re discussing edge routers? I don’t know enough about about the design of some routers to be able to properly comment. I can see why software is needed to strip the two bytes off the front, but with a trick or two this can be ameliorated, say putting the start of the buffer at address would n*16-14 then after just readjusting the buffer start point to be n*16+16, the remaining ‘buffer sans PPP’ is 16-byte aligned and there’s zero copying. But if your i/o hardware is able to split all the L2 L3 L4 headers apart and place them in separate small well-aligned buffers, then with PPP you’re losing that hardware capability, which is important. Being able to parse/split the protocol layers apart in an existing buffer in RAM by hardware would fix the problem though. But basically saying there’s no hardware support for PPP means that you’ve just chosen the wrong hardware, no?
-
Most hardware is built on commodity equipment. Firebrick are an exception however this is also why they still don't have an LNS released that can handle more than a gigabit.
The commodity hardware is knit together via regular north and south bridges with WiFi on a USB or PCIE bus and a routing ASIC purchased in bulk.
Some of these accelerate PPP, some don't. When PPP can be replaced it definitely should be, DHCP options and 802.1ad make it redundant.
-
> 40G a PC with an accelerator card, and a switch with a bunch of 40G ports.
Is that a fast commodity PC acting as a firewall-router?
-
No, the Pensando DSC-100 Distributed Services Card in it is doing the business. The PC doesn't have to be especially fast.
-
Here is a link to the product brief of the Pensando DSC-100 (https://pensando.io/wp-content/uploads/2020/03/DSC-100-ProductBrief-v06.pdf). (A PDF document.)
-
Wow that looks like some piece of kit. I would like to read some detail. It seems to be a hardware and software three ring circus.
-
Wow that looks like some piece of kit. I would like to read some detail. It seems to be a hardware and software three ring circus.
If you start from here (https://pensando.io/products/dsc/), you may find some further reading material.
-
https://youtu.be/_zSqft0cSfk might well be interesting.
-
Just for fun I punched the figures into Cisco's performance estimator (ngfwpe.cisco.com). With all firewall features enabled it recommends an FPR4125 which would be running at 70% capacity assuming default mix of packet sizes. There's a lot of high level security going on, so you could go a lot smaller and less expensive if you only want Layer 3 firewall.
For most of us I suspect 10gig to the home is so far away that current products aren't really relevant.
-
I wonder what is preventing 10G to the home. I suspect there will be little demand because of the limitations of WLANs for people who think that ‘wi-fi’ is the internet and the rest of us alike. But isn’t it true to say that once fibre is run in then the only thing preventing more speed is the PON hardware and replacing that is cheap and easy relatively speaking?
-
. . . isn’t it true to say that once fibre is run in then the only thing preventing more speed is the PON hardware and replacing that is cheap and easy relatively speaking?
Yes. The fibre has a massive bandwidth. So to upgrade just replace the optics in the OLT at the head-end and replace each ONT at the users' ends.
-
Ofcom / ASA regulations. Selling 10G over XGSPON isn't feasible for residential customers. It is only capable of 9.2 or so after overheads and any other usage on the PON will cause visible contention.
Elsewhere in the world this is okay as services may be sold at 'up to' but this isn't the case in the UK so we'll probably have to wait for 25G or higher before 10 is a residential product.
I may be able to purchase 10G over XGSPON in the near future but don't think even I can justify it :D
-
Plus if they roll out straight to XGSPON it could be contended with up to 64 users vs GPON 32.
I wonder if Openreach will switch to XGSPON once a PON reaches its limit as presumably "technically" a PON might cover more than 30 customers and there is an assumption they wont all go on FTTP any time soon? Or are they literally laying it where a PON will only ever cover 30 customers so XGSPON would have less contention than other countries rollouts?
Even I can't see a huge benefit in over a Gigabit any time soon, though XGSPON makes symmetrical more practical.
Are CityFibre using GPON or XGSPON given their service IS symmetrical?
-
A symmetrical service would be a big draw for me, given that backups are a nightmare.
-
There's also backhaul and ISP peering to consider. If an exchange had say 100 - 1000 10Gig customers, how much backhaul would the ISP need to keep those customers happy? Then at the ISP end multiply that by say 100-100 exchanges, how much Internet capacity would be needed, also I guess DNS and content caching performance.
-
Plus if they roll out straight to XGSPON it could be contended with up to 64 users vs GPON 32.
Some networks run GPON on a 128:1 split.
I wonder if Openreach will switch to XGSPON once a PON reaches its limit as presumably "technically" a PON might cover more than 30 customers and there is an assumption they wont all go on FTTP any time soon?
Very unlikely any time soon. On the 1000/115 product Openreach only guarantee the provider 110Mb/s.
It would take a lot to saturate a PON beyond that.
Are CityFibre using GPON or XGSPON given their service IS symmetrical?
Same as Openreach. GPON at 32:1
-
Plus if they roll out straight to XGSPON it could be contended with up to 64 users vs GPON 32.
I wonder if Openreach will switch to XGSPON once a PON reaches its limit as presumably "technically" a PON might cover more than 30 customers and there is an assumption they wont all go on FTTP any time soon? Or are they literally laying it where a PON will only ever cover 30 customers so XGSPON would have less contention than other countries rollouts?
Even I can't see a huge benefit in over a Gigabit any time soon, though XGSPON makes symmetrical more practical.
Are CityFibre using GPON or XGSPON given their service IS symmetrical?
GPON goes to 128 but it's a bad idea. You've just sliced another 6 dBm from the optical receive power at the ONT and increased loss, forcing transmit power to go higher. With the way the networks are built it adds complication hanging 4 x 32 port splitters off a 4
Openreach have shown no signs of being interested in XGSPON overlay. They have no XGSPON ONTs, and no interest in offering symmetrical even when the deploy XGSPON. New build sites are fibre only, so 30-32 customers per split.
CityFibre have both GPON and XGSPON on their network however the XGSPON tends to be used for mobile backhaul rather than delivery to end customer premises. The GPON is just accepting there may be some visible contention.
-
There's also backhaul and ISP peering to consider. If an exchange had say 100 - 1000 10Gig customers, how much backhaul would the ISP need to keep those customers happy?
A pair of 40G links would be ample both for capacity and resilience - should run fine even with just one of the 40G in place. Source: the 10/25G ISPs being able to run 25G services with dual 100G backhaul - quadruple your largest customer's capacity per OLT seems to be the sweet spot especially as it's so difficult to consume that much bandwidth.
Then at the ISP end multiply that by say 100-100 exchanges, how much Internet capacity would be needed, also I guess DNS and content caching performance.
At this point you have a CDN in the exchange next to your edge router. Sky already have CDNs in the largest exchanges and this number will only grow as time goes on.
-
GPON goes to 128 but it's a bad idea. You've just sliced another 6 dBm from the optical receive power at the ONT and increased loss, forcing transmit power to go higher. With the way the networks are built it adds complication hanging 4 x 32 port splitters off a 4
Openreach have shown no signs of being interested in XGSPON overlay. They have no XGSPON ONTs, and no interest in offering symmetrical even when the deploy XGSPON. New build sites are fibre only, so 30-32 customers per split.
CityFibre have both GPON and XGSPON on their network however the XGSPON tends to be used for mobile backhaul rather than delivery to end customer premises. The GPON is just accepting there may be some visible contention.
Yes I just had a quick read that a 32 split can go for 20km but 64 drops it to 5km. Plus as fibre is relatively cheap, it makes no sense to add limitations that could bite you in the ass years down the line if you run short of fibre to offer future speed increases, depending on what the technology requires. What you DON'T want is to have to actually add more fibres later and the insane civils costs that entails.
I know Openreach aren't interested in XGSPON or symmetrical today, but I think its inevitable eventually. This thread is after all mostly a thought experiment, even I wouldn't go for 10Gbit as the cost would be huge with little to no real-world benefits. I'm already quite happy with download speeds over Three 5G, the only problem is its unreliability vs a hard line.
-
Openreach have shown no signs of being inerested in XGSPON overlay. They have no XGSPON ONTs, and no interest in offering symmetrical even when the deploy XGSPON. New build sites are fibre only, so 30-32 customers per split.
To be fair to Openreach, they have looked at 10 and 25 XGSPON and are trialling Symmetric FTTP speeds (for business customers)
-
The XGSPON trial closed to new customers in January of 2021.
https://www.openreach.co.uk/cpportal/updates/briefings/ultrafast/nga203020
A gentle reminder to folks that GPON is 2008 technology, so both the line cards and ONTs are relatively really cheap, and that before 10G over XGSPON or close to it could be contemplated CPs would need access to higher capacity cablelinks. XGSPON at the headend is fine, the ONTs with 10G ports are a fair bit more expensive than the tiny GPON ONTs Openreach provide. I would bet those give change from £20 per unit.
-
The XGS-PON trial is indeed closed and was only available to order at specific locations for business customers.
In the end it was trialled at a single site.
-
In happier news other than Openreach and CityFibre most others seem to be using XGSPON by default now.
CityFibre do have XGSPON around, though. They use it to provide mobile backhaul in some cases.
-
Considering the huge size of the Openreach coverage area compared to alt-nets, it does make a lot of sense to go for the cheaper technology and upgrade to XGSPON later as and when its necessary.
The only reason I can see it as necessary is if contention kicks in on GPON, as I can't see Gigabit seeming too little for quite some time. I doubt most people will go for the higher packages anyway.
-
What he said ^^^^^^.
It's all well and good having 1Gbps symmetrical - but you also have to pay for that privilege - I'm happy with my 50Mbps FTTC circuit, satisfies our family needs.
-
For domestic users hooked on wi-fi there is a limit to what wireless LANs can provide and exceeding gigabit throughput seems to me to be unachievable because without new frequencies being allocated there is not the bandwidth available for very wide channels, not even if you’re prepared to be a very greedy anti-social hog eg 160 MHz 5GHz user.
-
For domestic users hooked on wi-fi there is a limit to what wireless LANs can provide and exceeding gigabit throughput seems to me to be unachievable because without new frequencies being allocated there is not the bandwidth available for very wide channels, not even if you’re prepared to be a very greedy anti-social hog eg 160 MHz 5GHz user.
Such as Wifi 7? https://en.m.wikipedia.org/wiki/IEEE_802.11be
-
For domestic users hooked on wi-fi there is a limit to what wireless LANs can provide and exceeding gigabit throughput seems to me to be unachievable because without new frequencies being allocated there is not the bandwidth available for very wide channels, not even if you’re prepared to be a very greedy anti-social hog eg 160 MHz 5GHz user.
Oh I do better than that. I have 160Mhz on my WiFi, and I had 80Mhz on a point to point link concurrently. :p
It didn't really impact anyone at the time as nobody had their routers on DFS channels. That's starting to become more common now.
Its annoying Apple didn't include 6Ghz in their latest Macbooks as I had planned to move over to WiFi 6e as soon as Zyxel release an AP for it.
To be fair, I don't really use WiFi much other than watching YouTube in bed sometimes but its a nice to have for OS updates or updating games on my gaming laptop, which I haven't really used since the pandemic as I'm not going anywhere.
-
I can’t get 5 GHz channels above 100 on my old ZyXEL WAPs. Is that where dynamic frequency selection lives?
Is my problem because of some stupid 30 min delay scanning for radar? Making me thing it’s stuffed when it’s just effect of the huge delay? Or am I just imagining such a thing? - half-understood from misreading something. I wonder if channel 100 will really work given sufficient patience. Not that waiting 30 mins, if true, would be practical.
I don’t have any near neighbours so I could use 80 MHz or 160 MHz with the right APs as long as I keep the power modest. My WAPs are nearly 12 years old now and I need to budget for replacement hardware before they just die on me. I have two working WAPs and also (I think) a third WAP in storage as a backup swap out unit so we are well-covered currently in case of hardware failure.
-
5Ghz DFS in the UK are channels 50 - 112, 116, 132 - 144.
-
I can’t get 5 GHz channels above 100 on my old ZyXEL WAPs. Is that where dynamic frequency selection lives?
Yeah its similar to why Zyxel wont enable Band C on their WiFi 6 APs, it requires going through re-certification to add the band, though I believe its even more awkward with Band B where DFS lives so it took many AGES to release units supporting it.
-
On the topic I will speak to YouFibre about a month trial of their 10G, strictly for test purposes ;) and see what happens.
-
On the topic I will speak to YouFibre about a month trial of their 10G, strictly for test purposes ;) and see what happens.
I'd be interested to see reults of that for peak, off peak and other stuff
-
Its funny as I've been "helping" someone over on Linus Tech Tips forum try to understand why his 5Gbit FTTP is being provided on a router with only 1x2.5Gbit and 2xGigabit ethernet ports, and why its really not worth his time trying to combine them.
I'm lucky to get 5Gbit between my NAS and desktop PC (which are connected at 10Gbit) because even between NVMe SSDs there seems to be bottlenecks on SMB and NFS.
-
What are those SSDs, 6Gbps ? can’t remember.
-
It depends what drives you buy.
My oldest NVME drive (Samsung SM951) "only" does 2,200MB/s read and 1,600MB/s write sequential.
Real world is much slower.
Some NVME drive are much slower than that, others are much faster. It depends if it's PCIE Gen 3 or 4, and how many lanes. It depends on how good the controller on the drive is.
-
Real world is much slower.
I inherited a small oldish Dell notebook/laptop recently without it's "hard drive" an (NVMe) SSD.
I bought the cheapest I could, a WD Blue SN550, which said it could achieve 2400MB/s.
Real world is at best 1500 read, 750 write, which is way more than I actually need - it's used for car diagnostics.
But I think we may be digressing rather too far from the original topic..... :-[
-
For heavy writes I've actually found SATA SSDs are better in my NAS.
Once SATA bottlenecks it doesn't hang system IO the same way NVMe does, I think because at 550MB/s the NAND keeps a constant flow of data whereas on NVMe it will burst at ungodly speeds to fill the DRAM cache and SLC cache, then IO hangs completely as the NVMe controller tries to move stuff into the MLC part of the NAND at double-digit speeds.
The end result is the NVMe probably completes faster but the computer is unusable during that time vs SATA where it continues to be responsive. Unless of course you have really REALLY high end NVMe drives where the cache is bigger than anything you're likely to write.
Found this out when extracting large rar/7z archives to the OS NVMe drive. Even when both the archives and extraction is to the same SATA SSD, it just works better than using the NVMe drive. Keep NVMe for small random reads where it excels.
Moving files over the network will never saturate either in my experience as you're adding more IO and network transfers are CPU heavy. Ironically I had to disable the NIC functions that reduce CPU load as they actually made things worse, especially on Windows clients.
-
Interesting, my SATA SSD systems tend to run at a symmetric (roughly) 500MB/s read and write (some really old ones are slower) but in the "real world" they're all about the same speed doing any task and that includes the NVMe ones.
Don't think I've got any rotary magnetic disks any more except for backups. Repurposing saves waste. :cool:
Ah, no, I do have one system on "traditional" hard disks. It's full of family photos and videos from about 1989 on.... Slaved to our biggest TV for nostalgia sessions.
-
Interesting, my SATA SSD systems tend to run at a symmetric (roughly) 500MB/s read and write (some really old ones are slower) but in the "real world" they're all about the same speed doing any task and that includes the NVMe ones.
Are you sure you mean NVMe and not M.2, as M.2 drives can be NVMe (PCIe) or SATA. No NVMe drive should be stuck at 500MB/s and they drop significantly lower once the cache is full. I have a fair selection of drives, most of my NVMe are fairly low end except my gaming PC which has a high-end Sabrent Rocket 4 Plus which is stupidly fast.
I hope Weaver doesn't mind that we've gone a little off-topic, though its relevant as one of the key factors in high-end FTTP not performing as expected is disk IO. This is another thing that comes up on the LTT forums quite regularly, people with Gigabit or faster but their CPU and/or SSD is too slow to keep up.
-
Quite sure, I've only got one NVMe system. Which is also M.2 and benchmarks quite fast.
The rest are just SATA replacements of magnetic hard drives. Typically Crucial BX500 range units.
I'm quite sure my systems are bottlenecked on CPU and/or RAM speed now.
-
Off-peak will drop once more people sign up as it's selling the full capacity of the PON however I don't expect significant drops.
I don't expect it to drop below 8, or go much above 9.
The main machine it'll connect to is a beast that should be more than able to handle the load.
-
@Alex no you’re fine - topic police have not been called ;) And in fact wasn’t I myself a contributor to ‘drift’?
If users’ CPUs, o/s’s and internal i/o hardware are not up to 10Gbps then will users even know? Average users perhaps don’t benchmark such things do they? Do some machines now have rubbish CPUs given the huge popularity of laptops and their low-power consumption requirements or thermal limitations. I would never have a laptop if it didn’t have to be portable - I hate them; sluggish performance, small pokey screens and awful abnormal keyboards. I will always want a huge PC with massive CPU oomph, proper keyboard and one or two giant monitors.
It could be that a significant use-case for 10Gbps will be multiple users though, I wonder.
-
If users’ CPUs, o/s’s and internal i/o hardware are not up to 10Gbps then will users even know?
ThinkBroadband is full of posts from users complaining they can't even max out the 900Mb/s package from their provider.
It almost always turns out to be their own hardware. Either a slow CPU, a rubbish NIC, or anti virus/security software showing their machine down.
Anyone who buys 10Gb/s FTTP would be doing so to spread that connection over multiple machines.
Either that or they need 1 hell of a system with drives setup in some form of RAID to get anywhere near that kind of throughput.
-
Some M.2 drives can handle 1.2 GB/s write. The 980 Pro in my primary machine can do over 5 GB/s in bursts and sustain 1.7 GB/s.
It's packing an Intel chipset NIC, so that should take most of the load off the CPU, and the CPU itself is Ryzen 59xx series.
Will see what it can do.
-
The 980 Pro and the Sabrent Rocket 4 Plus I have in my gaming PC are supposed to be fairly comparable. I can pull content form the NAS at 522MB/s to that machine no problem (from the SATA SSD, so basically maxing out), its sending to the NAS which tends to be the issue.
Copying from the Sabrent to the Intel 660p (high-end to low-end NVMe SSD) in that PC can do 2.1GB/s for a 3GB file but copying to the NAS with the P1 (which is basically the same SSD just different brand) bursts to 2Gbit or so then drops down to barely above Gigabit, though I just double checked and this seems to apply across all drives in the NAS so appears to be a different issue to what I was originally discussing (bottlenecking on copies entirely on the NAS itself).
This is doubly weird as the NAS has 25GiB of RAM assigned to cache right now so it "should" be able to just dump it into RAM at full speed then write it out to permanent storage in the background. Reading back that file does 560MB/s as its probably coming from that RAM cache so happily maxing out the 5Gbit link back to the main switch.
Its not the end of the world, its just rather odd and highlights how networking can be such a minefield of issues. Its never as simple as plug and play, very much the old MEME of plug & pray.
-
The last main workstation PC I had was a Dell with a 2 x SATA RAID SSD set up using a card that had a processor and lots of RAM on it.
I agree about the use-case for multiple PCs.
-
Every cabled device at home is connected directly to a switch with a 10G uplink so hopefully minimal bottlenecking.
I do have some work to do to improve it, though. Have the hardware ready to go so no rush.
-
I'm waiting on stock for the Netgear MS510TXUP to add more 10Gbit ports to the core network and so the main switch can become the media/console switch with multi-gig and fibre uplink instead of copper. Plus I wanted SNMP on the second switch so I can monitor traffic better.
If I had a network closet I would have gone with a used enterprise switch from the start, but everything is on a shelf in the hallway so can't be too whiny or you'd hear it in every room.
Unfortunately I upgraded mum to a NUC so the only chance of getting her above 1Gbit would be expensive Thunderbolt adapters. I tried USB which works, but they take too long to come up on boot and break the boot-time NFS mounts. Plus I'm using the single USB-C port for the monitor as its so much neater than thick Displayport cables.
-
Have tons of 10G ports but cabling is a challenge so the more I can send down a single cable the better.
-
Can you aggregate some of your 10G pipes into one cable perhaps, or is that not workable ? Or LACP say?
-
LACP requires multiple physical links to be used, as would a port channel of multiple links. Need multiple links going into a switch with an SFP28 port and that can then talk to the existing SFP28 at the other side.
Or I purchase new switches on both, leaving 20 unused ports each side and making 2 8 port switches redundant.
Or I move some equipment around and change the path so it's not the current 'tromboning' and then upgrade if spine capacity becomes an issue though it shouldn't do with any regularity. Simultaneously using the full Internet and on that same switch in the same direction transferring between there and NAS isn't going to be common.
-
Selling 10G over XGSPON isn't feasible for residential customers. It is only capable of 9.2 or so after overheads and any other usage on the PON will cause visible contention.
Was wrong here - it's actually about 8.5G over XGSPON.
https://youtu.be/VK1Tt6v-vWU
-
Was wrong here - it's actually about 8.5G over XGSPON.
:swoon: Yeah the whole point is that if you have the capacity on the back-haul anyway, might as well let the customers use it than have artificial profile caps and spare capacity being unused.
[Moderator edited to remove the unnecessary link to the YouTube video.]
-
That happens plenty on both XGSPON and GPON, a single customer being sold all the capacity and having an 'up to' speed of the full PON.
Can't be done in the UK due to advertising regulations.
-
Can't be done in the UK due to advertising regulations.
Seems to be what a lot of the alt-nets are doing for upstream though, as presumably thats not covered by the advertising regulations?
-
Nearly. The gigabit over GPON services leave a couple of hundred Mb available to other customers. Seem to be enough customers in lightly loaded areas to be able to do this and hit the numbers.
Selling uncapped the ISP has to manage allocation of capacity more and unless they've extensive data be very cautious with the advertising.
-
A lot of my data usage is off-peak hours so it tends to work well, much better I'd expect on FTTP rather than mobile services which are just all over the place in the day time and periodically go mental during peak hours (200ms+ spikes when idle). Even 5G doesn't seem to have fixed that, presumably due to not being standalone mode yet and upstream having a harder time reaching the mast due to the obstructions I have.
It should certainly be interesting once I have the choice of OR and CityFibre, but I suspect thats another year or two away for the latter as they don't seem to be moving very quickly (presumably as they have to install cabinets so they seem to have only done areas with existing ones for now vs OR which do not have that issue).
Even so, watching one.network feels like OR are moving at a glacial pace too. I suspect once all ducts are cleared things will seem to move a whole lot faster.
-
There's a flurry of activity which then dies down.
If you have a look at somewhere like Brighton you can see that burst.
Sheffield has a bunch of activity from VM lined up which is positive.
It's a good sign when Openreach duct clearance dies down. Should indicate they are cabling and splicing.
-
Yes, I would not be surprised if VM roll out FTTP too here as they curiously did some work this end of the street that was never cabled. Could be that with OR and CityFibre covering it soon and duct sharing being presumably easier for fibre, that they finally see the appeal.
-
Could be that with OR and CityFibre covering it soon and duct sharing being presumably easier for fibre, that they finally see the appeal.
It's happened elsewhere!