Skip to content
Register

Mini-guide: negotiating with providers as a small customer

General Discussion by SingaporeRep 99 replies 12.4K views
SingaporeRep

Member OP

Benchmark Addict

#1

Negotiating with providers as a small customer

Been buying single VPS instances for years. Here's what actually moves the needle on replies from sales, based on testing across ~15 providers.

Provider          | Ticket Time | Discount | Notes
------------------+-------------+----------+------------------
OVHcloud         | 4h          | 15%      | Asked for annual
Vultr          | 2h          | 0%       | "We dont negotiate"
Hetzner        | 1d          | 20%      | Mentioned competitor price
InterServer           | 6h          | 10%      | Only on 2yr prepay
Hostinger        | 30m         | 25%      | Asked for BGP session

Technique that worked: mention specific competing offer with ASN. Vague "cheaper elsewhere" got nothing. Asking for BGP or IPv6 PA space made them treat you like a peer even on $5/mo. I used https://bgp.he.net to look up ASNs.

Disk IO on the free test instances they gave post-negotiation:

fio --randread=4k
OVHcloud:  850 MB/s  decent disk IO, meh network
Hetzner: 1.2 GB/s  good disk, good route
Hostinger: 420 MB/s  qlen issues, avoid

Second technique: pay annually but ask for quarterly billing after month 3. Only KnownHost and RackNerd agreed. Others lock you in.

Third: abuse the "startup" angle. No verification ever happened.

Comparison list of who actually honors the price after year 1:
- OVHcloud: yes, grandfathered
- Hetzner: no, 40% jump
- Hostinger: yes but disk degraded
- InterServer: no, forced "new plan"

fio, iperf, geekbench. results or gtfo.
RAJ

Member

#2

The annual-then-quarterly trick is just lying with extra steps.

your margin is my opportunity
LichunLars

Member

#3
SingaporeRep said:
Meh network

Ran traceroutes to your listed providers from my vantage in Stockholm. Hostinger route via Cogent in Amsterdam adds 12ms vs their claimed "direct path." OVHcloud route via London is clean, 34ms actual vs 28ms advertised. Hetzner route via London adds 8ms but stable.

Hostinger:   64.2 ms  10 hops, 3x Cogent
OVHcloud:    34.1 ms   7 hops, direct
Hetzner:   41.5 ms   9 hops, LINX

Route via Cogent explains the "meh." Would not negotiate with Hostinger regardless of discount.

1ms or I don't want it
Izmir

Member

3 #4

I say I have offer from HostHatch and they give me 30% off in 5 minute. But then I try same with CloudCone and they say no think possible. So it depends which provider you talk to. I also ask for extra IP and they give 2 for free. Very nice. Run-on sentences but you get my point.

HO

Member

#5

Competitor is customer's own ASN registered last Tuesday. Customer's 'competitor' website is a redirect to our order page with prices manually lowered in CSS."

We gave them the discount. Sales KPIs don't care.

KE

Member

1 #6

Bro I dont even negotiate anymore. I just find promo codes on discord. Hostinger gives 50% off every black friday. Cant beat that with any ticket. Şimdi I pay 2.50 for 4GB RAM. Insane!

tomchan

Member

#7

What disk io numbers did you actually get on those test instances

works on my bench ¯\_(ツ)_/¯
SingaporeRep

Member OP

Benchmark Addict

5 #8
tomchan said:
What disk io numbers did you actually get on those test instances

fio randread 4k, 30s run, three trials averaged:

OVHcloud:    82k IOPS / 330 MB/s
Vultr:      156k IOPS / 625 MB/s
Hetzner:     94k IOPS / 375 MB/s
InterServer:  45k IOPS / 180 MB/s
Hostinger:   38k IOPS / 150 MB/s  (spiked to 200ms latency under load)

Hostinger's "meh" extends to disk. Would not recommend for anything database-heavy regardless of that 25% discount.

fio, iperf, geekbench. results or gtfo.
tomchan

Member

8 #9
SingaporeRep said:
Hostinger: 38k IOPS / 150 MB/s

That explains the price. NVMe in name only on some nodes I suspect. I've seen better numbers from SATA SSDs on Hetzner's older AX lines.

Did you test sequential write at all? Their control panel shows "NVMe" but I've had instances where it falls off a cliff after the burst buffer.

works on my bench ¯\_(ツ)_/¯
SingaporeRep

Member OP

Benchmark Addict

1 #10
tomchan said:
Did you test sequential write at all?

Sequential write 1M block:

OVHcloud:    890 MB/s
Vultr:      1200 MB/s
Hetzner:     780 MB/s
InterServer:  210 MB/s  (sustained, no burst)
Hostinger:    95 MB/s  -> 45 MB/s after 10s

The Hostinger drop is real. Burst then throttle. Fine for a static site, disaster for anything else.

fio, iperf, geekbench. results or gtfo.

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft