5 Gbps guaranteed bandwidth

At 5Gbps, the Disk Runs Out Before the Port Does

Guaranteed 5Gbps across twelve locations, and a chassis choice that decides whether cold reads can keep up with the port you are paying for.

X-ZoneServers 5 Gbps dedicated servers start at €519 per month for a guaranteed 5 Gbps port — 7 bare-metal configurations with up to 1.62 PB of monthly transfer across 12 EU and US datacenter locations. The port is effectively unmetered: hold the full committed rate around the clock, with no overage charges and no fair-use throttling inside your tier.

519
starting price per month
103.8
per Gbps per month
7
configurations
1.62 PB
transfer ceiling per month

Most capacity questions stop at the port. This one should not. A guaranteed 5Gbps link asks your machine for 625 MB/s of payload every second, for as long as the traffic lasts, and that figure sits squarely in the range where the storage subsystem starts to matter more than the network card. If the bytes are already in memory, any of the seven builds on this page will hold the line without effort. If the bytes are on disk and the requests arrive in no particular order, a six-bay SATA SSD chassis and a ten or twenty-four bay NVMe chassis behave like different products. That difference, not the port rating, is what you are really choosing between here.

What the Port Asks For, and Who Has to Answer It

Held flat across a 30-day month, this tier moves 1.62 PB. A single terabyte crosses the wire in 26 minutes and 40 seconds. In streaming terms it supports 1000 concurrent 1080p viewers at 5 Mbps, or 200 concurrent 4K viewers at 25 Mbps. Read those three figures as demands on your array rather than as headline ceilings, because the network side of each is identical and the storage side is not. A thousand viewers are a thousand independent read positions in your library, and unless the popular titles are resident in page cache, most of those positions are a seek. The terabyte transfer is the gentle case: it is one long sequential read, and every array here handles sequential reads well. The 4K sessions sit in between, fewer streams but much larger blocks and a working set that grows faster than memory does. The uplink delivers all three the same way. Your disks will not.

1.62 PB
Transfer per month
Sustained volume at full port utilisation across a 30-day month.
1,000
1080p viewers
Simultaneous streams at 5 Mbps each.
200
4K viewers
Simultaneous streams at 25 Mbps each.
26 min 40 s
Per terabyte
Wall-clock time to move 1 TB off the server at line rate.

Figures use decimal units (1 Gbps = 125 MB/s) across a 30-day month, at 5 Mbps per 1080p stream and 25 Mbps per 4K stream. They describe what the port can carry, not a guarantee of application throughput: disk, CPU and the far end all apply their own limits.

7 configurations at 5Gbps

Three of the seven builds share one storage layout: six SATA SSD slots. The dual Xeon E5-2630v3 at €519 brings 16 cores and 32 threads with 128GB of DDR4-2133. The dual Xeon E5-2690v4 at €539, our popular pick, brings 28 cores and 56 threads on the same 128GB. The dual Xeon E5-2699v4 at €609 brings 44 cores and 88 threads with 256GB. Note that the extra cores buy you no extra bays. The EPYC side changes the geometry entirely. The single EPYC 9254 at €769 with 24 cores and 48 threads, and the dual EPYC 9254 at €999 with 48 cores and 96 threads, both carry ten NVMe slots backed by DDR5-4800. The dual EPYC 9554 at €1279 carries 24 NVMe slots and 512GB, and the dual EPYC 9754 at €1839, our ultimate configuration, carries 24 slots and a full 1024GB. Choose on bays and memory first, cores second.

Dual Xeon E5-2630v3

Entry Level

519
/month
16c / 32t @ 2.4GHz128GB DDR4 @ 2133MHz
6x SSD Slots5Gbps Guaranteed
AI Optimized Routing Included
Order Now
POPULAR

Dual Xeon E5-2690v4

Popular Choice

539
/month
28c / 56t @ 2.6GHz128GB DDR4 @ 2133MHz
6x SSD Slots5Gbps Guaranteed
AI Optimized Routing Included
Order Now

Dual Xeon E5-2699v4

High Performance

609
/month
44c / 88t @ 2.2GHz256GB DDR4 @ 2133MHz
6x SSD Slots5Gbps Guaranteed
AI Optimized Routing Included
Order Now

Single EPYC 9254

Entry Level

769
/month
24c / 48t @ 2.9GHz128GB DDR5 @ 4800MHz
10x NVMe Slots5Gbps Guaranteed
AI Optimized Routing Included
Order Now

Dual EPYC 9254

High Performance

999
/month
48c / 96t @ 2.9GHz256GB DDR5 @ 4800MHz
10x NVMe Slots5Gbps Guaranteed
AI Optimized Routing Included
Order Now

Dual EPYC 9554

Ultimate Performance

1279
/month
128c / 256t @ 3.1GHz512GB DDR5 @ 4800MHz
24x NVMe Slots5Gbps Guaranteed
AI Optimized Routing Included
Order Now
ULTIMATE

Dual EPYC 9754

Maximum Performance

1839
/month
256c / 512t @ 2.3GHz1024GB DDR5 @ 4800MHz
24x NVMe Slots5Gbps Guaranteed
AI Optimized Routing Included
Order Now

What a 5Gbps server is for

Long-Tail Video Libraries

A catalogue larger than memory turns every unpopular title into a cold read, and the tail is exactly where requests scatter. Ten NVMe bays absorb that scatter. Six SATA bays begin to queue behind it.

Restore-Heavy Backup Targets

Ingest is sequential and easy. Restores are not. Several customers pulling different archives at once is random read at full rate, and that is precisely where bay count and queue depth decide the result.

Patch Day Asset Distribution

A new build pushes one hot working set at everybody simultaneously. Either it fits in memory, and the entry chassis holds the port comfortably, or it does not, and NVMe is the difference between line rate and a disk-bound plateau.

Public-Facing Read Replicas

Serving query results to the open internet mixes small random reads with occasional large scans. Once the index outgrows cache, sustained delivery becomes a property of the storage layout rather than of the uplink.

The Price of the Port, and the Price of Delivering It

Entry is €519 per month, which works out at €103.8 per Gbps of guaranteed port. That is the number to start from, but it is not the number to decide on. If your working set fits inside the entry chassis's memory, €519 buys the whole tier and nothing is stranded. If it does not, the comparison that actually matters is €519 against €769 for the single EPYC 9254, the cheapest build here with NVMe bays, because that €250 is what turns a port rating into a delivered rate on cold data. Everything above it in the range buys cores, memory and bays rather than a faster link, since all seven configurations carry the same 5Gbps commitment with no overage charges inside the tier. So budget in two parts. Decide the port first, because it is fixed and cheap to reason about. Then decide how much of it you intend to serve from disk, and let that set the chassis. Buying the port twice over is the expensive mistake, and it is the one nobody notices until the traffic is real.

All guaranteed bandwidth tiers

The same hardware range is sold at nine committed port speeds. Compare the entry price of each.

5Gbps dedicated server questions

Can six SATA SSDs actually serve this tier?
Sequentially, yes and easily. A striped six-drive SATA array reads large contiguous blocks well past the payload rate this tier requires. The difficulty is random access. Each SATA device works from a single shallow command queue, so once requests arrive out of order and in small sizes, the array spends more of its time waiting on command turnaround than transferring data. Whether that ever becomes visible depends entirely on how much of your traffic is answered from memory rather than from the drives.
How much RAM do I need before the disk stops mattering?
Enough to hold the working set, not the whole dataset. The page cache keeps recently read blocks in memory, so a very large archive with a hot region of a few tens of gigabytes behaves like a memory-resident workload even on the 128GB entry build. Measure that hot region under real traffic rather than estimating it, leave room for the application itself, then compare against the 128GB, 256GB, 512GB and 1024GB options in the line-up. If the hot region moves around unpredictably, hit rates fall and the bays do the work instead.
Which RAID layout gives the best read throughput?
Striping is what scales reads, and mirrored stripes give you throughput plus a rebuild that does not have to read every surviving drive. Parity layouts read perfectly well when the access pattern is sequential, but small random reads punish them, and a degraded array punishes them further at exactly the moment you need full rate. On the NVMe builds many operators skip hardware parity altogether and let the filesystem or the application handle redundancy, because the drives stopped being the constraint some time ago.
Why might the cheapest chassis not sustain the tier from disk?
Turn the payload rate into operations and it becomes clear. At a 64 KB average object size it is close to 10,000 reads per second, and at 4 KB it is over 150,000. A six-bay SATA array delivers that from cache without trouble, and delivers it from disk without trouble when the reads are sequential. What it cannot reliably do is that rate as cold random reads while writes are also landing. The entry chassis is not underspecified, it is specialised: it holds the tier when your access pattern is friendly to it.
Do filesystem and readahead settings change anything here?
They change more than most people expect. Readahead set too small leaves NVMe queues half empty on streaming reads, so the drives never reach the rate they are capable of. Readahead set too large burns bandwidth fetching blocks nobody requested and evicts genuinely useful pages from cache. Match it to your average object size, keep filesystem allocation groups spread across the array so writes do not concentrate on one device, and mount with noatime so read traffic stops generating writes. Our own administrators will tune this with you at handover if you ask.

5 Gbps dedicated servers by location

Same hardware and guaranteed port, deployed from the datacenter closest to your users.

Tell Us the Access Pattern, Not Just the Traffic

Send us your working set size, average object size and read mix, and we will tell you which of the seven builds sustains 5Gbps from disk rather than only from cache. Custom CPU, memory to 1TB DDR5, storage and OS quoted on request, with deployment in about an hour.

Up to 1Tbps DDoS protectionDeployed in about 1 hour99.9% uptime SLA