When I upload an archive, I know that uploading price is null, but requests gone in more than 200 000, so my estimated bill is something 15-20 $, so I get one way, I put it my all files to a zip or tar (Compress), & is there any alternate way to reduce my requests to upload archive ?
first, glacier is built for archives, i.e for retaining data on a very long term, where retrievals are infrequent and unusual. It should not be accessed unless very rare occasion. Think about Glacier as a warehouse where you would pack up your data and consign them.
So you should not do a lot of requests to Glacier, and if so you should consider using S3 instead.
That been said, you should avoid to zip/tar all your files in one archive file, as you can retrieve up to 5% of your average monthly storage for free each month.
However, keep in mind that each archive object has 32 kb of data overhead associated with it. Thus, an archive size of 3.2 MB would have only 1% of cost overhead, when a 1KB archive would have 97% of cost overhead.
So if you have too small files, aggregate them in larger files; if you have too large files, split them in smaller ones. you just have to find the good balance to keep your archives easy enough to browse, e.g using archive descriptions or a local index (e.g nosql store), or both.