Hi, Yunfeng
Yunfeng Wu <wyfsky...@126.com> 于2020年8月18日周二 下午8:11写道: > > > > Can we provide some performance performance comparision between v1 and v2. I was worried about the perfomance would slow down when do some string > field aggs opertion because of the `look up` of the `dict ordinary` > > Actually, in theory, segment v2 is more efficient for reading strings than v1. Because it stores encoded numbers, lighter IO operations are required when reading. As for the Look-up operation you mentioned, it is an operation of O1 in the memory, and the time cost is negligible. I will appreciate if you can provide some performance comparison test reports. Thanks, Zhao Chun > > > > > > > > -- > > ------------------------------------------ > 武云峰 > email:wyfsky...@126.com > > > > > > 在 2020-08-18 17:22:22,"Zhao Chun" <zh...@apache.org> 写道: > >+1 > > > > > >Thanks, > >Zhao Chun > > > > > >41108453 <41108...@qq.com> 于2020年8月18日周二 下午12:24写道: > > > >> Yes, it should be like this, saving you having to configure each time > >> > >> > >> > >> > >> ------------------ 原始邮件 ------------------ > >> 发件人: > >> "dev" > >> < > >> morning...@163.com>; > >> 发送时间: 2020年8月18日(星期二) 中午12:14 > >> 收件人: "doris-dev"<dev@doris.apache.org>; > >> > >> 主题: [DISCUSS] Set SegmentV2 as the default format of Doris Storage > >> > >> > >> > >> Hi All: > >> > >> > >> I suggest that in version 0.13, SegmentV2 is set as the default storage > >> format. > >> All newly created tables are in V2 format. > >> > >> > >> The table created by the user before still retains the original format, > >> and the user can choose to convert it by himself. > >> > >> > >> > >> -- > >> > >> 此致!Best Regards > >> 陈明雨 Mingyu Chen > >> > >> Email: > >> chenmin...@apache.org >