All indexers of Kepler need to fill out the following form
This data is to help us understand the costs being incurred by each indexer so we can better understand the profitability and guide future consumers to decide on pricing.
---- Please copy and fill in below ----
What cloud provider are you using (or did you run your own hardware)
What dictionaries are you indexing (Polkadot, Kusama, Both)
If running both, did you run this in a single VM, or did you run all the services in seperate instances
Compute:
What type of compute are you using (e.g AWS ec2 t3a.large)
What has the average/max CPU been
What is your best estimate for your weekly compute cost per project (e.g. just Polkadot)
Storage/DB:
What type of storage are you using (SSD, HDD, RDS/Database service)?
What is your best estimate for your weekly storage cost per project (e.g. just Polkadot)
Data egress:
How much data egress do you see each week
What is your best estimate for your weekly egress cost per project (e.g. just Polkadot)
Total cost
What is your best estimate for your weekly total cost per project (e.g. just Polkadot)
What other docs or guides can we write to help you
What monitoring tools or data would you like us to provide
What cloud provider are you using (or did you run your own hardware)
Hetzner
What dictionaries are you indexing (Polkadot, Kusama, Both)
Both
If running both, did you run this in a single VM, or did you run all the services in seperate instances
Separate docker containers (single dedicated server)
Compute:
What type of compute are you using (e.g AWS ec2 t3a.large)
Dedicated bare metal server
What has the average/max CPU been
1 cpu 8 cores
What is your best estimate for your weekly compute cost per project (e.g. just Polkadot)
I don’t know
Storage/DB:
What type of storage are you using (SSD, HDD, RDS/Database service)?
SSD NVMe RAID 1 drives
What is your best estimate for your weekly storage cost per project (e.g. just Polkadot)
I don’t know
Data egress:
How much data egress do you see each week
300Gb per week income traffic
What is your best estimate for your weekly egress cost per project (e.g. just Polkadot)
I don’t know
Total cost
What is your best estimate for your weekly total cost per project (e.g. just Polkadot)
100 euros
What other docs or guides can we write to help you
N/A
What monitoring tools or data would you like us to provide
Grafana monitoring
• What cloud provider are you using :Hetzner
• What dictionaries are you indexing : Both
• If running both, did you run this in a single VM, or did you run all the services in seperate instances : Single
• Compute:
o What type of compute are you using (e.g AWS ec2 t3a.large)
o What has the average/max CPU been 1 cpu 8 cores
o What is your best estimate for your weekly compute cost per project (e.g. just Polkadot)
• Storage/DB:
o What type of storage are you using (SSD, HDD, RDS/Database service)? NVME
o What is your best estimate for your weekly storage cost per project (e.g. just Polkadot) $25
• Data egress:
o How much data egress do you see each week 70G
o What is your best estimate for your weekly egress cost per project (e.g. just Polkadot) $3
• Total cost
o What is your best estimate for your weekly total cost per project (e.g. just Polkadot) $25
• What other docs or guides can we write to help you NA
• What monitoring tools or data would you like us to provide Prometheus/Grafana
• Is there anything else that we can do better? NA
What cloud provider are you using (or did you run your own hardware) -Hetzner
What dictionaries are you indexing (Polkadot, Kusama, Both) -Both
If running both, did you run this in a single VM, or did you run all the services in seperate instances Separate docker container on dedicated server
Compute:
What type of compute are you using (e.g AWS ec2 t3a.large) AMD Ryzen 5 3600 6-Core Processor
What has the average/max CPU been 12% max -30%
What is your best estimate for your weekly compute cost per project (e.g. just Polkadot) Im here to validate, not to calculate
Storage/DB:
What type of storage are you using (SSD, HDD, RDS/Database service)? NVME RAID 0
What is your best estimate for your weekly storage cost per project (e.g. just Polkadot) Im here to validate, not to calculate
Data egress:
How much data egress do you see each week 250-270GB income traff for 1 week
What is your best estimate Im here to validate, not to calculate
Total cost
What is your best estimate for your weekly total cost per project (e.g. just Polkadot) 500 buks ^)
What other docs or guides can we write to help you Only rules. can i make one more validator?
What monitoring tools or data would you like us to provide Yes, we need a system to keep track of whether our node is alive or not. Whether it’s doing well or not, how many blocks per second it can process and how many requests per second it can process. Is everything okay with her right now? I had cases when the traffic was 10Mbit and when it was 1.2Mbit and in both cases the node looks alive, but how should it be?
What cloud provider are you using (or did you run your own hardware)
Contabo
What dictionaries are you indexing (Polkadot, Kusama, Both)
Both
If running both, did you run this in a single VM, or did you run all the services in seperate instances
run in single VM
Compute:
What type of compute are you using (e.g AWS ec2 t3a.large)
Cloud VPS with SSD Storage VPS XL SSD: 10 cores 60 GB 1,600 GB (100% SSD) 1,000 Mbit/s
What has the average/max CPU been
15%/100%
What is your best estimate for your weekly compute cost per project (e.g. just Polkadot)
N/A
Storage/DB:
What type of storage are you using (SSD, HDD, RDS/Database service)?
SSD
What is your best estimate for your weekly storage cost per project (e.g. just Polkadot)
N/A
Data egress:
How much data egress do you see each week
N/A
I’m using 32 TB Out + Unlimited In
What is your best estimate for your weekly egress cost per project (e.g. just Polkadot)
N/A
Total cost
What is your best estimate for your weekly total cost per project (e.g. just Polkadot)
$15/week . 65$/month
To meet 2 projects I use about 600GB SSD, VM has 6 cores, 16G Ram.
If running 3 projects, the capacity will be about 800GB.
I’m using the Monthly Plan, about $65/month, excluding Onfinality’s $50 API fee.
The total cost of running Node Kepler costs about $115/month
What other docs or guides can we write to help you
N/a
What monitoring tools or data would you like us to provide
N/A
Is there anything else that we can do better?
Multi Dictionary in Project, to reduce the load on indexers, and save money on buying RPC.
What cloud provider are you using (or did you run your own hardware)
Hetzner
What dictionaries are you indexing (Polkadot, Kusama, Both)
Both
If running both, did you run this in a single VM, or did you run all the services in seperate instances
dedicated server, both containers on it
Compute:
What type of compute are you using (e.g AWS ec2 t3a.large)
12core CPU, 128gb RAM, 2TB NVMe raid0
What has the average/max CPU been
10%/30-40%. I’ve thought that would be 3 projects to index, not 2.
What is your best estimate for your weekly compute cost per project (e.g. just Polkadot)
~50 euros for both, so it’s 25 per project
Storage/DB:
What type of storage are you using (SSD, HDD, RDS/Database service)?
2tb NVMe RAID0
What is your best estimate for your weekly storage cost per project (e.g. just Polkadot)
Hard to calculate, N/A
Data egress:
How much data egress do you see each week
Over 200GB AFAIK
What is your best estimate for your weekly egress cost per project (e.g. just Polkadot)
50 euros that I’ve answered above - it’s about all costs, can’t calculate for each thread
Total cost
What is your best estimate for your weekly total cost per project (e.g. just Polkadot)
Oh I see It’s 50 euros
What other docs or guides can we write to help you
I gues more accurate info about how networks works, how any numbers are calculating, also what we can expect about Kepler (new projects to index?), something like that
What monitoring tools or data would you like us to provide
N/A
Is there anything else that we can do better?
You’re doing great, I hope some kind of stress tests will come to us!
What cloud provider are you using (or did you run your own hardware)
Contabo
What dictionaries are you indexing (Polkadot, Kusama, Both)
Both
If running both, did you run this in a single VM, or did you run all the services in seperate instances
single VM
Compute:
What type of compute are you using (e.g AWS ec2 t3a.large)
VPS: 10 cores 60 GB RAM 1600GB (100% SSD) 1,000 Mbit/s
What has the average/max CPU been
40%/100%
What is your best estimate for your weekly compute cost per project (e.g. just Polkadot)
N/A
Storage/DB:
What type of storage are you using (SSD, HDD, RDS/Database service)?
SSD
What is your best estimate for your weekly storage cost per project (e.g. just Polkadot)
N/A
Data egress:
How much data egress do you see each week
N/A
32 TB Out + Unlimited In
What is your best estimate for your weekly egress cost per project (e.g. just Polkadot)
N/A
Total cost
What is your best estimate for your weekly total cost per project (e.g. just Polkadot)
~30$ week both projects until I have enough disc space
What other docs or guides can we write to help you
I need new SSL configuration guide. I have a technical background, but I haven’t worked with nginx and ssl. The current guide seems to be missing some important points or some points are meant to be obvious, so if you follow it step by step, you will not be able to set up the ssl.
What monitoring tools or data would you like us to provide
N/A
Is there anything else that we can do better?
I see two problems:
Documentation quality:
When I ran the indexer for the first time, the documentation suggested taking a preconfigured image, which I did, but nothing worked. I asked a question on the forum, and it turned out that this is an old image that has not been updated for a long time and should not be used.
Recently, a monitoring guide was published in the chat, in which there was a command that replaced the docker-compose file.
When the dictionary files were replaced, the old hash sums remained in the documentation.
The academy has a section for indexers and it indicates where to get information about all updates of the coordinator and proxy - the link leads to the Notion and if you click it you will see “access is denied”.
etc. I can go on with the list.
Good documentation is a separate big task. Those who already know what and how to do don’t need documentation, so we need to write it with the expectation that a person is doing it for the first time and it is necessary to describe in detail all the nuances and it is better to try this guide to make sure that there are no mistakes in it.
The second problem is a continuation of the previous one. I see a lack of communication. If something went wrong and we encountered some problems, then we ask questions in the chat and on the forum. And some questions go unanswered for weeks. You ask a question, then someone else asks a question, the discussion moves on and no one returns to the old questions.
I propose the following solutions:
Make a one-page guide for indexers that always contains the latest carefully tested working version.
As an option for those instructions that most likely will not change over time (installing the SSL for example) make a video guide. If the author misses something important in the text, users will always be able to see it on the video and will not make mistakes.
Implement a ticket system for indexers in Discord. So that no problem is left unattended. And accordingly, we need a team member who would regularly check tickets and answer questions.
What cloud provider are you using (or did you run your own hardware)
I run two indexers on two separate vps. I use my own hardware at the moment (2x32core XEON, 128Gb RAM, 4x1Tb 15k hdd on raid 10).
What dictionaries are you indexing (Polkadot, Kusama, Both)
Both.
If running both, did you run this in a single VM, or did you run all the services in seperate instances
Single VM.
Compute:
What type of compute are you using (e.g AWS ec2 t3a.large)
VM 16 cpu, 16 Gb.
What has the average/max CPU been
8/48 cpu and 16/48 Gb. I’ve started indexing on 8cpu/16Gb and add cpu and memory. The system has entered a stable state with 16cpu/16Gb. Further addition of resources did not give any improvements.
What is your best estimate for your weekly compute cost per project (e.g. just Polkadot)
I suppose 70-80 eur monthly cost for server (17-20 eur per project).
Storage/DB:
What type of storage are you using (SSD, HDD, RDS/Database service)?
RAID10 hdd.
What is your best estimate for your weekly storage cost per project (e.g. just Polkadot)
I suppose 70-80 eur monthly cost for server (17-20 eur per project).
What cloud provider are you using (or did you run your own hardware) Hetzner
What dictionaries are you indexing (Polkadot, Kusama, Both) Both
If running both, did you run this in a single VM, or did you run all the services in seperate instances Single
Compute:
What type of compute are you using (e.g AWS ec2 t3a.large) AMD Ryzen 7 PRO 1700X 8-Core 64 GB
What has the average/max CPU been 5% / 25%
Storage/DB:
What type of storage are you using (SSD, HDD, RDS/Database service)? 1TB SSD
Data egress:
How much data egress do you see each week ~400k responses for both projects
Total cost
What is your best estimate for your weekly total cost per project (e.g. just Polkadot) 30$ per project, 60$ for both.
Is there anything else that we can do better?
I have to agree with some of the previous statements that the documentation quality could be improved. I believe the instructions should be more detailed, and it seems that some of them were not adequately tested.
There is technically no Discord moderation, which leads to a lack of structure and communication. Many questions remain unanswered, and looking into channel history doesn’t always provide help. There is no definitive information on the current official version of the coordinator and proxy, resulting in confusion, and often upgrading leads to further mistakes. The last pinned message is dated 29.04, while it should be updated with the most common mistakes, the current official working version, and other relevant information.
It would be helpful to have an active moderator in the indexers channel.
