Time for the seasonal Solana roast review

Its a meme that Solana in its high speed transactions does tend to fail a lot a transactions. But it might not be entirely Solana’s fault.

Task

• Create a dashboard displaying Solana network performance over time.
• How has the network performed over the past month compared to the rest of the year?
• Has transaction per second and success rates of transactions gone up recently?
• Is this because of less botting or fewer users, or new improvements from the Solana engineers?
• What wallets and programs have paid the most in fees for failed transactions?

\n Method

Multiple tables from Flipside’s Data were used:

• For success/fails ratios, flipside_prod_db.solana.fact_transactions table was used.

  • Also to account for their fees.
  • The same with distinct users.

• To account for a more complete view of transactions per second (TPS),

  solana.core.fact_blocks table was used.

• To look for the top programs, solana.core.fact_events.program_id table was used.

  • And to label them, solana.core.dim_labels.

• And to find Bots, solana.core.fact_swaps table was used

I swear its only the fail/success ratio

From the transactions per second, that can be calculated strictly from the transactions table, we can see that a non minor share of them results in failure.

And well, curiously, the amount of fails does not correlate with network usage, in fact, we can see that even when activity is minimal, there still are a lot of failing transactions.

Sadly, the transactions table, the one where is straightforward to see failed ones, doesn’t include all the data flowing through the blocks table. That would be the rightest most chart.

We can see that recently Solana activity it has been spiking on the blocks table though.

Yes, it fails, but what else this last month?

With 434 millions in failed transactions cumulating 2380 SOL just in fees. Almost half the transactions on Solana

The succeeded ones amounted a total of 2380 in Sol fees.

And well, there was less users than April but with less changes.

I mean, it distinct users where in fact more in April, but, after that fall in amount of users, numbers didn’t go back to what they were before.

They are receding though, just not back to early 2022 levels.

And even then, the average daily fee in failed transactions in June was 79 Sols

More than the previous months with similar activity, it even has less transactions than January & February but has a bigger fail rate.

And frankly im not sure why, but i have a theory. But you gonna have to keep reading

> ## I swear this has never happened to me before.

                                (Solana after its quintillion failed transaction)

Its the bots, has to be the bots

I mean, they have not stop existing, but there are fewer than before.

To search for them the amount of each one participating in swaps was count for and divided buy hour, then grouped by it gives us a view of at least on specific group, the high frequency trading swapping ones.

Those bots are the ones executing large amount of swaps orders. But there is something in particular here, while i couldn’t see this directly on flipside’s data, i did find that some of the failed transactions are registered as cancelled by the users.

This bots are making high swapping orders and cancelling them. A chunk of the transactions doesn’t necessarily fail because the network fails, because it does, but instead they are marked as fails because they were cancelled.

And in June?

Bot activity doesnt appear to be very present in June, well, actually by this lense, no month would appear to be botted, but at least in June the amount of non high frequency swapers looks high.

Now, which programs did fail the most?

Mango and Jupiter, by so much its not even a competition

And makes sense, Jupiter is an aggregator of multiple dexes and Mango allows for making long and short trades. Making both of them particularly attractive to botting.

Can we locate this big losers?

In amount of failing transactions, of course.

Yes and no.

Of course we can just take these big payers in any Solana explorer, but can we go further?

Probably, but not me.

After I found these high fees payers I cross searched them, they are not programs.

And only a few of them participate in swapping.

By order, only the 3rd, 6th, 8th, 9th and 10th Biggest fees contributors are participating in high frequency swapping.

> And oh boy do they swap.

Last words

June was a particularly weird month.

• A lot of failed transactions.

• Not so many users.

• Fewer bots

• But also higher daily fees.

  \