With　the　development　of　cloud　computing,　the　requirements　for　dynamic　distribution　of　cloud　computing　according　to　the　real-time　load　are　getting　higher　and　higher.　It　is　necessary　to　predict　the　load　of　cloud　computing　resources　to　achieve　the　goal　of　real-time　elastic　scaling.　Cloud　computing　load　will　constantly　change　with　time　and　actual　demand.　Therefore,　lstm-attention　model　is　used　to　collect　data　and　model　and　predict　cloud　computing　resource　load　for　a　long　time,　so　as　to　achieve　accurate　prediction　of　cloud　computing　resource　load　in　the　future.It　can　be　concluded　that　the　lstm-attention　model　performs　well　in　cloud　computing　resource　load　prediction.　©　2020　Institute　of　Physics　Publishing.　All　rights　reserved.